Geomorphically dynamic arid landscapes in western NSW are the focus of this research on Aboriginal stone artefact scatters.

Geomorphology, Archaeology and Geoarchaeology:

Introduction and Background Chapter 1: Introduction

CHAPTER ONE

GEOMORPHOLOGY, ARCHAEOLOGY AND GEOARCHAEOLOGY: INTRODUCTION AND BACKGROUND

1.1 Thesis Aims and Scope

This thesis explores the interrelationships between geomorphology and Aboriginal archaeology in western New South Wales (NSW), Australia, in particular how geomorphic processes at a variety of scales influence the spatial and temporal distribution of the surface stone artefact record that archaeologists study. It also explores how aspects of the archaeological record can inform on past landscape change. A major outcome of the research is a new geoarchaeological approach to surface artefact survey and interpretation anchored by an understanding of geomorphic landscape dynamics.

Australian Aboriginal archaeology has, for most of its history, focused on stratified deposits found in rock shelters, caves and overhangs (Robins 1996), where traditional archaeological techniques of excavation and salvage have been employed (e.g. Allen 1996; Dortch 1979; Hale & Tindale 1930; Mulvaney & Joyce 1965; O’Connor 1995; Roberts et al. 1990; Smith 1987). This is in spite of the fact that surface scatters of stone artefacts, together with associated hearth remains, are the most ubiquitous form of Aboriginal archaeological record across the whole of the Australian continent (Hiscock & Hughes 1983; Holdaway & Stern, in press). Whilst there are many reasons for this discrepancy, a key issue is the lack of understanding of the interrelationships between the stone artefacts discarded by Aboriginal people in prehistory and the geomorphic processes that occur on the land surfaces upon which they are found, such that the surface archaeological record has often been dismissed as hopelessly contaminated in terms of trying to understand hunter-gatherer behaviour (Robins 1997). Because they are so common, surface artefact scatters are the mainstay of much field archaeology, particularly that connected with environmental impact assessment, and are the subject of many of the archaeological conservation decisions made in Australia (HOLDAWAY ET AL. 1998). However, few Australian studies have investigated the level of disturbance of surface artefact scatters (Robins 1993) and none have examined the impact of geomorphic landscape dynamics on the archaeological record of Aboriginal hunter- gatherer activity.

1 Chapter 1: Introduction

That is not to say that archaeologists have ignored or disregarded geomorphology and geomorphic processes. On the contrary, they have long been recognized as part of the suite of 'natural' formation processes that, together with cultural processes, determine the shape of the archaeological record (e.g. Nash & Petraglia 1987; Schiffer 1983, 1987; Wood & Johnson 1978). Geologic and geomorphic input is essential to interpreting the stratigraphic sequences of ‘sites’, especially the nature of the environments of deposition that are represented by the archaeological sediments. Indeed, a whole sub-discipline of archaeology - geoarchaeology - has grown out of these kinds of applications of standard geologic and geomorphic techniques (e.g. Rapp & Gifford 1995, Rapp & Hill 1998; Waters 1992).

In Australia, the interrelationship between geomorphology and Aboriginal archaeology has been dominated by the investigation of Late Quaternary palaeoenvironments, most notably the work of Bowler (1970, 1971, 1973, 1976) at Lake Mungo (Figure 1.1), in tandem with archaeological investigations that seek to determine the time of earliest colonization of the Australian continent by Aboriginal peoples (e.g. Bowler et al. 1970, 1972; David et al. 1997; Fullagar et al. 1996; O’Connell & Allen 1998; Thorley 1998b; Thorne et al. 1999; Turney et al. 2001). Related to this is the study of people/environment interactions, in particular the ways in which Aboriginal people adapted to the variety of environmental settings in which they lived (Holdaway & Stern, in press). Once the antiquity of Aboriginal occupation of the Australian continent was realized, prehistoric settlement patterns in relation to the distribution of resources such as water, and how this had changed over the tens of thousands of years of Aboriginal occupation, became a focus for research (e.g. Ross 1984; Ross et al. 1992; Smith 1989; Thorley 2001; Veth 1993). Conversely, evidence for modification of the environment by Aborigines has also been a focus for archaeological research (e.g. Hughes & Sullivan 1981; Jones 1968, 1969; Mulvaney 1969).

Not surprisingly, in view of the lesser attention given to surface artefact scatters, landscape- based archaeological studies are relatively few. Early work includes that of McBryde (1968, 1974) in northeast NSW and Hallam (1977) in southwest Western Australia (WA). More recently, landscape-based approaches have been attempted by archaeologists trying to understand regional patterning in artefact assemblages (e.g. Robins 1993; Thorley 1998a; Witter 1992). However, to date, there have been no published Australian studies where geomorphologists and archaeologists have collaborated from the outset to fully integrate

2 Chapter 1: Introduction geomorphological and archaeological survey and analysis methods to investigate spatial and temporal patterns of prehistoric Aboriginal activity.

Figure 1: Australia showing the Stud Creek location and other places mentioned in the text. The arid zone, defined by the 250 mm annual rainfall isohyet, is indicated by hatching, and carets show the general location of the Great Dividing Range.

The research presented here fills that gap. It focuses on the landscape settings of surface artefact scatters, but utilizes the record of landscape change preserved in the landforms and underlying sediments to provide a temporal framework for those scatters. It also takes advantage of the fact that, in western NSW, exposure of artefact scatters is high. Large numbers of artefacts can be surveyed across many thousands of square metres without disturbing the remaining archaeological record by excavation, satisfying the desires of Aboriginal custodians for their cultural heritage to remain in place. Recent advances in digital technology have meant that a permanent record of the artefactual material, within its landscape context, can be made and retained, like a traditional museum collection, for future research.

3 Chapter 1: Introduction The specific aims of the research are to: 1. investigate recent landscape change in western NSW and the contemporary geomorphic setting of surface artefact scatters; 2. determine the degree of disturbance of those scatters by geomorphic processes and whether it affects their potential for informing on prehistoric Aboriginal ‘use of place’; 3. develop a chronology for landscape history and Aboriginal occupation in the study area using absolute and relative dating of both the sedimentary and the archaeological record; 4. present a geoarchaeological framework for surface artefact survey and analysis that takes account of their contemporary geomorphic landscape setting and the history of landscape change.

1.2 Thesis Structure

The thesis is organised into five chapters. Chapter One sets out the aims and scope of the study, and the structure of the thesis. The origins and evolution of geoarchaeology are then reviewed, with particular emphasis on the development of landscape-based approaches to understanding and analysing the archaeological record, both internationally and in Australia. The background to the current research is then outlined. Two published papers that conclude the Chapter describe the results of a pilot study where several of the initial hypotheses were tested, and outline the scope of the research program.

Chapter Two presents geomorphic evidence for recent landscape change in western NSW that has resulted in the exposure of surface scatters of Aboriginal stone artefacts across extensive areas. The notion of a dynamic landscape setting for surface artefact scatters is established, and the significance for archaeological research outlined.

Chapter Three describes the artefact survey protocols developed to accommodate this dynamic landscape setting, and presents the results of analyses of the spatial distribution of surface artefacts designed to establish their lateral integrity.

4 Chapter 1: Introduction Chapter Four presents a two-stage framework for establishing the chronology of surface artefact scatters in western NSW, using radiocarbon determinations from heat-retainer ovens as well as stratigraphic analysis and dating of valley fill sediments.

Chapter Five synthesises the outcomes of the research into a model of spatial and temporal variability in the archaeological record in western NSW. The veracity of the model is then tested using data from a pilot study at a different location within the region. The implications for current models of Holocene Aboriginal settlement patterns, and for various cultural heritage management issues, are discussed, and a new geoarchaeological framework for investigating surface artefact scatters that is anchored in an understanding of geomorphic landscape dynamics is proposed.

The bulk of the thesis comprises a series of papers, either published (7 papers), or submitted to a journal for consideration (1 paper). The introduction to each chapter summarises the respective papers, reviews additional relevant published material, and provides the context in terms of the overall aims of the thesis for the papers that it contains. Thus, the references for each chapter are listed at the end of that chapter. Papers which form part of this thesis are referenced in upper case throughout. Spelling and terminology follow Australian English standards, except where publication has required American English conventions.

1.3 Geoarchaeology: Origins and Evolution

Northern Hemisphere Perspectives Geoarchaeology is generally regarded to be the application of the geosciences to solve research problems in archaeology (Butzer 1982; Pollard 1999). While the term was first used relatively recently (Renfrew in Davidson & Shackley 1976), interaction between the geosciences and archaeology goes back to the early nineteenth century, when geology and prehistoric archaeology developed essentially in parallel (Pollard 1999). Other terms such as archaeogeology, archaeological geology and archaeometry have also been used in the same context, and although the differences in meaning between them are considered to be trivial (Herz & Garrison 1998), there appears to be a general consensus that geoarchaeology is particularly concerned with geomorphology, pedology, stratigraphy, sedimentology and chronology (e.g. Gladfelter 1977; Pollard 1999). Rapp & Hill (1998) distinguish geoarchaeology from archaeological geology by considering the former to be part of archaeology itself, for example, analysis of the stratigraphy of archaeological sites, while

5 Chapter 1: Introduction the latter is research that is essentially geological in nature but that has implications for archaeology, for example, studies of coastal landforms and sea level change in the Mediterranean that had important repercussions for interpretation of the development of human societies and trade in the region.

Substantial advances in the knowledge of the earlier stages in human evolution, particularly in Africa, were the result of co-operative work between archaeologists, geologists and biological anthropologists, amongst others (Harris 1980). Koobi Fora in northern Kenya and Olduvai Gorge in Tanzania are “…good examples of well-investigated locations within sedimentary lake basins, where hominid remains and archaeological sites occur in a variety of [landscape] settings…” (Harris 1980: 63). In particular, advances in isometric dating techniques, notably potassium/argon age determinations on volcanic ash and lava, palaeomagnetic stratigraphy, and fission-track dating, provided the means by which environmental change as well as biological and cultural development at these locations could be assessed.

Geoarchaeology as a sub-discipline of both archaeology and geology is more formally recognised in the U.S.A. than elsewhere (Herz & Garrison 1998). Most of the monographs on the subject have been published in North America (e.g. Rapp & Gifford 1985; Rapp & Hill 1998; Waters 1992), as is the journal Geoarchaeology, and there are specialist geoarchaeology subgroups of both the Geological Society of America and the Society for American Archaeology (Goldberg et al. 2001). The impetus was the research focus of New World archaeology on when and how humans first colonised the American continents (Pollard 1999). The ephemeral nature of the archaeology of crucial early palaeoindian sites created the necessity for close collaboration between archaeologists, geomorphologists and sedimentologists, whose skills have been essential to understanding the archaeological record (Pollard 1999). The general impression of the nature of geoarchaeological study, as soil or sediment analyses from stratigraphic sequences providing palaeoenvironmental information and relative dating for the archaeological material they contain (Butzer 1982), was cemented by this research.

However, Gladfelter (1981) argued that, rather than taking a secondary role, geoarchaeological involvement should occur at all stages of archaeological investigations, i.e. design, excavation and analysis, and include geophysical exploration techniques, identification of the spatial context of sites, differentiation of natural and cultural formation

6 Chapter 1: Introduction processes in site formation, development of temporal contexts by absolute and relative dating, and reconstruction of palaeo-landscapes. In accordance with this view, a surge in studies of ‘natural’ (i.e. geomorphic and pedologic) site formation processes in the 1970s and 1980s helped to broaden the scope of geoarchaeology, although it mostly grew out of the desire by archaeologists to infer behaviour from artefacts (e.g. Schiffer 1972). To do this, the post-depositional effects of both natural and cultural processes had to be identified and accounted for (Stein 2001). Interestingly, much of the experimental and observational work on artefact taphonomy and post-discard redistribution processes was done by archaeologists rather than geoscientists (e.g. Petraglia & Nash 1987; Rick 1976; Shackley 1978; Schick 1987; Stein 1983; Villa 1982; Villa & Courtin 1983; Wandsnider 1989).

In the last two decades, there has been a proliferation of new approaches in archaeology that can be considered geoarchaeological in nature, particularly the expansion of spatial data recovery and analysis away from traditional focus on specific locations in the landscape, or archaeological ‘sites’, to the incorporation of distributional and non-site data from across extensive regions (Rossignol & Wandsnider 1992). Geomorphological analysis of landscapes has become increasingly important in analyzing archaeological materials and understanding the shape of the archaeological record (e.g. Bettis & Mandel 2002; Buck et al. 1999; Doleman 1992; Doleman & Stauber 1992; Doleman et al. 1992; Kuehn 1993; Seaman et al. 1988; Wandsnider 1989; Zvelebil et al. 1992). However, only a few of these studies combined geomorphological and archaeological survey and analysis techniques from the outset. The geomorphological dynamics of the landscape were more often examined post-hoc in order to explain the shape of the archaeological record. However, three recent publications (Wells 2001; Barton et al. 2002; Bettis & Mandel 2002) illustrate the importance of a fully integrated geoarchaeological framework for analyzing human land use and settlement patterns in the past.

Using case studies from coastal Peru and Cyprus, Wells describes “…methods by which geomorphology can be integrated into an archaeological survey to facilitate survey sampling strategies, prioritize survey regions, reconstruct palaeolandscapes, and provide an environmental framework for survey data interpretation…” (Wells 2001: 108). Echoing Gladfelter (1981) twenty years previously, she emphasizes the importance of geomorphologists being involved in survey project planning from the outset so that landscape analysis can provide the basis for sampling strategies. Subdivision of landsurfaces on the basis of relative stability aids in the determination of which surfaces are most likely

7 Chapter 1: Introduction to have archaeological material exposed at the surface, buried beneath sediments, or completely eroded away. In the Cyprus study, a geographic information system (GIS) was used to statistically compare the landscape classification maps with artefact distribution and density. A chronology provided by the relative dating of geomorphic surfaces allowed determination of which parts of the landscape were extant during any particular period of occupation. Thus, the method allowed for a spatial stratification of the landscape based on the highest likelihood for artefact discovery, resulting in the most efficient use of valuable field time.

Barton et al. (2002) have gone further and used these techniques to develop a diachronic model of land use change over the 80,000 years of human occupation of the Polop Alto valley in eastern Spain. A series of maps depicting spatial patterning in landuse over time were generated in a GIS by combining artefactual with landscape and stratigraphic data, and interpreted in terms of frequency, duration, density and area of occupation to construct a picture of ‘use of place’ by human inhabitants of the valley since the Paleolithic. Recognition of the dynamic nature of the geomorphic landscape assisted in the modeling of the dynamics of human use.

Finally, Bettis & Mandel (2002) summarise the controls of spatial and temporal patterns of fluvial system activity on the preservation and visibility of the archaeological record of past human activity in the central and eastern Great Plains of the U.S.A. Using a method of analysis reflecting earlier work of Waters (1991, 2000) and Waters & Kuehn (1996) in the American southwest, Bettis & Mandel (2002) summarise alluvial stratigraphies from selected river basins reflecting the range of landscapes across the west/east environmental gradient from south-central Kansas to central Iowa, and look for patterns in the record of alluvial sedimentation that may inform on preservation of the archaeological record in those basins. They found that periods of aggradation and channel erosion were diachronous throughout drainage networks across the region, and hence the Holocene sedimentary record is not uniformly preserved. This in turn determines the degree of preservation of cultural deposits dating to particular periods in different locations within the drainage basins. They attribute the relative abundance of Late Prehistoric sites to geomorphic conditions which favoured their preservation and visibility, and reject the notion that the greater number of sites dating to the late Holocene reflects dramatic population increases.

8 Chapter 1: Introduction Geoarchaeology in Australia Like ‘New World’ archaeology in North America, Australian archaeology has been largely dominated by the quest to find the site of earliest colonization of the Australian continent and to understand how and when Aboriginal people first arrived here. This work has always had a strong interdisciplinary flavour, with geologists, geomorphologists and pedologists making important site-specific contributions to a number of ‘classic’ studies of the pioneering period of archaeological research in Australia (Hughes and Sullivan 1982; Shawcross & Kaye 1980). However, geoarchaeology has had little recognition as a distinct sub-discipline of either archaeology or geology in Australian universities, with most archaeology being taught in Arts rather than in Science faculties. Nevertheless, many archaeological studies have been conducted in Australia that incorporate geological and geomorphological investigations of archaeological materials and their landscape settings.

Hughes and Sullivan (1982) published a review of Australian geoarchaeology in the proceedings of the first Australasian archaeometry conference held at the Australian Museum in January 1982 (Ambrose & Duerden 1982). However, the majority of papers at this and later conferences focused on the geoscientific applications to archaeology more commonly recognized as archaeometry, such as the chemistry and provenance of artefacts and archaeological sediments, and conservation of archaeological materials (e.g. Ambrose & Mummery 1987; Fankhauser & Bird 1993; Prescott 1988). One exception was a paper by Williams (1982), in which he used examples from the Nile Valley in Africa, the River Son in north-central India, and the Shaws Creek rockshelter in eastern Australia to emphasise a geomorphologist’s view that prehistorians need to take careful account of landscape change when interpreting archaeological deposits.

Hughes and Sullivan (1982) identified two main groupings of Australian geoarchaeological studies: those drawing on the methods and theories of geomorphology, geology and pedology which have been largely site-specific in approach, and those drawing additionally on the geographical sciences which have tended to be regional in focus, utilising a wide range of environmental and spatial approaches. From the first group, they reviewed investigations of the stratigraphy and chronology of rockshelter sites, the nature and sources of raw materials for stone artefact manufacture, and rock art conservation studies. Two further areas of investigation, namely palaeoenvironmental and chronological investigations of ‘open sites’ like Lake Mungo and river terraces in Victoria, and the impact of Pleistocene and Holocene environmental changes, particularly sea level change, on the pattern of

9 Chapter 1: Introduction Aboriginal settlement along the southeastern Australian coastline, are mentioned but not reviewed.

Writing in 1982, Hughes and Sullivan indicate that landscape-based archaeological research was just gaining interest in Australia at that time. Worrall (1980) set out to investigate the theme of “man…as an agent of geomorphic change” in the Mangrove Creek catchment in coastal NSW (Figure 1.1), but instead discovered that geomorphic dynamics had significant impacts on preservation of the archaeological record of human activity. Unfortunately, this important insight was never pursued. Most early attempts at landscape-based archaeology in Australia focused on developing models of prehistoric settlement patterns based on the interpretation of sites in their Holocene paleoenvironmental context (e.g. Bonhomme 1983; Ross 1981, 1984; Smith 1988). However, in contrast to the serendipitous approaches of earlier archaeological research in Australia, these researchers used systematic surveys to look for sites, with stratified random sampling based on geomorphic criteria. More recently, Witter (1992) undertook a study of regional variation in stone artefact assemblages across NSW using geographic criteria, including Land Systems classification, to try to predict site locations and contents across a broad range of environments, from the humid Great Dividing Range to arid northwestern NSW (Figure 1.1). An objective of this study was to develop criteria upon which the assessment of site significance could be based for cultural heritage management purposes. However, the scale of the study was at odds with the nature of the stone artefact assemblages, with within-site variability overwhelming any regional trends. A predictive model never eventuated.

Robins’ (1993) study of the archaeology of the Currawinya Lakes region of southwest Queensland (Figure 1.1) was the first comprehensive attempt by an Australian archaeologist to test the utility of a landscape-based approach to identify and explain patterns of archaeological variability through time and space. Robins used non-site archaeological survey (after Thomas 1975) to identify spatial patterns in the archaeological record at a regional scale, and geomorphological survey, taphonomic experiments, and excavation and dating of several hearths and rockshelter deposits to try to understand those patterns. His stratified, systematic, transect-based sampling strategy was based on broad physiographic units (sandplains and dunefields versus dissected residuals) and land systems (Robins 1997). Regional availability of key resources (water and stone) was the major determinant of archaeological site patterning in the study area (Robins 1993, 1997).

10 Chapter 1: Introduction While recognizing at the outset that "…the archaeological story…will be inextricably bound up with the geomorphic history…" (Robins 1993: 7), only limited integration of geomorphology into the methodological and analytical framework was undertaken. A major limitation of the study was the lack of consideration of geomorphic landscape evolution over time scales commensurate with human occupation of the study area, particularly European, and its control of artefact exposure and visibility. Despite a reference (Robins 1993: 93) to a particular location being "…extensively modified by European development…", Robins appears to assume that the geomorphic landscape has changed little in the time since Aboriginal occupation of the region, and that the landscape characteristics observable today were also present to more or less the same degree when stone artefacts were manufactured, used and discarded. This is certainly not the case for western NSW (FANNING 1999) and is unlikely for southwestern Queensland, although detailed studies of recent geomorphic landscape evolution in this region have not been published. It will be demonstrated in Chapter Two of this thesis that the land use change that accompanied European occupation of western NSW resulted in widespread erosion of topsoils in some areas, deposition of sediments in others, and incision of formerly stable valley floors, and that these processes had significant effects on artefact exposure and visibility. Moreover, episodes of geomorphic landscape instability have been characteristic of all parts of the Australian arid zone throughout the Late Quaternary. Without survey and analysis methods that take account of these influences on the shape of the archaeological record, only very coarse associations between patterning in artefact scatters and human behaviour are possible.

A regional landscape-based approach was also used by Thorley (1998a) to investigate arid zone settlement patterns and human adjustment to environmental change. Whereas previous investigations had utilized isolated sites spanning the central Australian ranges and their hinterland (e.g. Smith 1989; Veth 1993), Thorley chose sites that were bounded within a single catchment. Intensive, systematic surveys that incorporated both archaeological and geomorphological data were carried out in three study areas that reflected different landscape characteristics within the Palmer River catchment of Central Australia (Figure 1.1). Thorley (1998b, 1999) expected to find differences in occupation history at the three locations based on differences in water permanency, landforms, and positioning in the catchment. But in spite of supposedly abundant resources, especially water, and the relatively large area sampled (60,000 m2), only 411 artefacts were found. While dismissing differential visibility as a factor, Thorley (1998a, 2001) acknowledges that sediment loss in

11 Chapter 1: Introduction the catchment has accelerated since the introduction of pastoralism in the late 1800s. However, his research neglects the geomorphic dynamics of the landscape and the effects it might have had on preservation of the archaeological record, and instead seeks explanation in ethnographic accounts of place use (Thorley 2001).

1.4 A New Landscape-based Geoarchaeological Approach to Australian Surface Artefact Scatters

Rossignol (1992: 4) defined a landscape-based approach as "…the archaeological investigation of past land use by means of a landscape perspective, combined with the conscious incorporation of regional geomorphology, actualistic studies (taphonomy, formation processes, ethnoarchaeology), and marked by on-going re-evaluation of concepts, methods and theory…" While regional geomorphology (e.g. Witter 1992) and studies of artefact taphonomy (e.g. Robins 1993) have provided a physical landscape context for studies of surface artefact scatters, and ethnography a social landscape context (e.g. Thorley 1998a), the most critical shortcoming of landscape-based archaeological research in Australia to date, as highlighted by the foregoing review, is the lack of Rossignol’s “re- evaluation of concepts, methods and theory”. In particular, Australian archaeologists have failed to recognise the impact of geomorphic landscape dynamics, at a range of scales, on the spatial and temporal patterning of the archaeological record, except in a fairly superficial sense through a few artefact taphonomic studies. As a consequence, they have continued to apply survey and analysis methods and to develop Aboriginal hunter-gatherer settlement models that assume spatial and temporal uniformity in the magnitude and frequency of landforming processes and, in some cases, stable landscapes since the time of formation of the archaeological record.

However, geomorphic processes are not homogeneous in operation over a wide variety of spatial scales (Chorley et al. 1984), and are temporally variable, particularly in arid environments. In Central Australia, for example, rainfall is 10 to 20 % more variable than the world average for comparable areas, and episodic events largely shape the environment (Friedel et al. 1990). Together with human-induced environmental change, this episodicity of landforming events has compounded the direction of geomorphic landscape evolution set by Late Quaternary climate change, as first documented by Bowler et al. (1976). Geomorphic landscapes show evidence of both functional (i.e. form – process) and

12 Chapter 1: Introduction historical influences, and are a palimpsest1 of relict and modern forms (Chorley et al. 1984), episodically modified by contemporary geomorphic processes. There are many parallels with archaeological landscapes, which are themselves highly variable in time and space: a palimpsest of artefacts surviving multiple behavioural events (e.g. Aston & Rowley 1974; Barker 1982). In fact, archaeological landscapes are a complex mosaic of landsurfaces of variable ages containing records of human activity of variable lengths. The key to understanding the archaeological record initially lies with understanding the dynamics of the geomorphic landscape in which it is found.

Wells (2001) and Barton et al. (2002) have recently demonstrated the importance of the integration of geomorphology in archaeological investigations in Cyprus and Spain. The research presented in this thesis adopts a similar approach to surface artefact scatters in arid Australia, independently developing a landscape-based framework for artefact survey and analysis anchored by an understanding of geomorphic landscape dynamics. Regional landform pattern is not simply used as a static basis for spatial sampling stratification (e.g. Robins 1993; Thorley 1998a; Witter 1992). Instead, the morphodynamics and geomorphic evolution of the landscape are used to focus artefact surveys on those parts of the landscape where the degree of exposure and preservation of the archaeological record are likely to be maximised. Whereas artefact taphonomic studies have been primarily used to prove post- discard disturbance, and hence devalue surface scatters as a suitable vehicle for archaeological research, here they are used to demonstrate the overall lateral integrity of those scatters. Differential artefact visibility at the time of archaeological survey can be viewed as a function of contemporary landscape processes, allowed for in the survey protocol.

Temporal frameworks need not depend on the relatively few reliable chronologies built from absolute dating of long stratigraphic sequences preserved in caves or rockshelters and extrapolated across the continent. Local chronologies can be developed from absolute and relative dating of geomorphic surfaces and the archaeological materials, such as hearths, that they preserve. Rather than seeking to date individual events, whether behavioural or geomorphological, the goal of landscape-based archaeological research should be to seek patterns in both the spatial and temporal record of events that will allow a detailed

1 Greek: palin – ‘again’, psao – ‘rubbed smooth’; ‘writing-material or manuscript on which the original writing has been effaced to make room for a second writing; monumental brass turned and re-engraved on the reverse side’ (The Concise Oxford English Dictionary, 6th edition. Clarendon: Oxford.) 13 Chapter 1: Introduction prehistory of Aboriginal ‘use of place’ (sensu Wandsnider 1989) to be developed. The Western New South Wales Archaeology Program was established to pursue that goal.

1.5 The Western New South Wales Archaeology Program and the TIB13 Pilot Study

Archaeologist, Dr Simon Holdaway, initiated the Western New South Wales Archaeology Program (WNSWAP) in 1995. A stone tool specialist with experience in applying Geographic Information Systems (GIS) to archaeological problems, his aim was to develop new techniques for examining the surface stone artefact record in ways that would inform on the spatial and temporal patterns of prehistoric Aboriginal hunter-gatherer activity. As indicated by the foregoing review, surface scatters had previously been largely dismissed by archaeologists because they were unbounded, lacked the stratigraphy considered essential for establishing a chronology of occupation, and appeared to be significantly disturbed by post-depositional formation processes.

Initial discussions with Dr Dan Witter, then Western Region archaeologist with the NSW National Parks and Wildlife Service, led to the selection of a pilot study area in Sturt National Park in far northwestern NSW that Witter had designated the ‘TIB-13 site’ (see Figure 1 in HOLDAWAY ET AL. 1998 for location map). As noted by Witter (1992: 142), the site had been ‘…heavily eroded and all of the artefactual debris are [sic] resting on a hard clay surface…’ This meant that the artefacts were easy to see and to survey, but erosion processes that had exposed them may have compromised their vertical and lateral integrity. The need to undertake the artefact survey within a geomorphological framework was obvious.

The first of two papers that comprise the rest of this chapter (HOLDAWAY ET AL. 1998) describes the results of that pilot study, and the second (HOLDAWAY ET AL. 1997) summarises the rationale for future directions for WNSWAP research. Our main aim in undertaking a pilot study was to investigate whether, by studying surface scatters of stone artefacts, we could detect variability in the spatial deposition of artefacts in the past. At the same time, we investigated the influence of post-discard geomorphic processes on those spatial patterns, and the suitability of newly developed rapid data capture technology and GIS for investigating problems of this type. As suggested by Wandsnider (1992), landscape-

14 Chapter 1: Introduction based distributional approaches to artefact survey requiring large volumes of feature and artefact attribute data have only been made possible by these technological advances.

The outcomes of the research confirmed that our approaches to tackling the problems associated with surface artefact scatters were sound, and warranted further development. There was a clear correlation between artefact density and landsurface type, most likely reflecting visibility differences across the study area. Artefacts located on or close to areas of concentrated water flow, such as rills, appeared to be size-sorted, but there was no clear relationship between artefact size and distribution pattern on lagged surfaces away from the rills. By limiting assemblage analysis to only those artefacts found on lagged surfaces, we detected significant spatial patterning in assemblage variability across the site (HOLDAWAY ET AL. 1998).

While the time frame for deposition of the artefacts at TIB-13 could only be inferred to be mid- to late Holocene, based on tool typology, there was sufficient geomorphic evidence to indicate that a chronology for landsurface change at this and nearby locations could be developed by sedimentological analysis and absolute dating of the valley fill sediments upon which the artefact scatters were now resting. This would provide a maximum age for deposition of the artefact scatters. At the same time, radiocarbon dating samples from the exposed remains of numerous heat-retainer ovens (also called ‘hearths’) associated with the artefact scatters had the potential to provide a chronology of occupation across the site, thereby answering the critics who considered surface scatters to be undateable because they lacked the stratigraphy usually required. The range of dates from the radiocarbon determinations would indicate an ‘envelope of time’ during which the associated artefacts were deposited.

We had the team and the technology (Figure 1.2); the money came in the form of an Australian Research Council Industry Collaborative Grant between our research team and the NSW National Parks and Wildlife Service for further research in Sturt National Park from 1996 to 1998. Entitled ‘Predictive Modeling of the Distribution of Archaeological Materials in Sturt National Park’, the project aimed to develop a generally applicable set of methods for archaeological research and heritage management by abandoning the archaeological concept of ‘sites’ and substituting a geomorphic landscape framework upon which to undertake artefact survey, analysis and assessment of scientific significance. The second paper in the chapter (HOLDAWAY ET AL. 1997) outlines the scope of this

15 Chapter 1: Introduction research. It was originally published on CD ROM and has been reformatted and condensed for reproduction here.

The research presented in the rest of the thesis largely draws on the outcomes of this research project. I focus on my geomorphic contribution, particularly in recognizing and documenting the processes of landscape evolution responsible for preservation of the archaeological record and its exposure at the surface, and the impacts of contemporary processes on differential visibility and lateral integrity of surface artefact scatters. I demonstrate how an understanding of geomorphic landscape history and dynamics is essential for building a chronology of Aboriginal occupation of the Australian arid zone.

16 Chapter 1: Introduction

Figure 1.2: The WNSWAP team of (l. to r.) Trish Fanning, Dan Witter and Simon Holdaway “demonstrating” some of the technology to visiting press. The eroded valley floor of Stud Creek, with the field crew analysing artefacts, can be seen in the background.

17 Chapter 1: Introduction

18 Chapter 1: Introduction

1.6 References

Allen, J. (1996, ed.). Report of the Southern Forests Archaeological Project. Vol. 1: Site Descriptions, Stratigraphies and Chronologies. Archaeology Publications: School of Archaeology, La Trobe University, Bundoora. 277 pp. Ambrose, W. & Duerden, P. (1982, eds). Archaeometry: an Australasian perspective. Department of Prehistory, Research School of Pacific Studies, Australian National University, Canberra. Ambrose, W. & Mummery, J.M.J. (1987, eds). Archaeometry: further Australasian studies. Department of Prehistory, Research School of Pacific Studies, Australian National University, Canberra. Aston, M. & Rowley, T. (1974). Landscape Archaeology: an Introduction to Fieldwork Techniques on Post-Roman Landscapes. David & Charles: London. Barker, P. (1982). Techniques of Archaeological Excavation. Batsford: London. 2nd edition. Barton, C.M., Bernabeu, J., Aura, J.E., Garcia, O. & La Roca, N. (2002). Dynamic landscapes, artefact taphonomy, and landuse modeling in the Western Mediterranean. Geoarchaeology: an International Journal 17(2), 155-190. Bettis, E.A. III & Mandel, R.D. (2002). The effects of temporal and spatial patterns of Holocene erosion and alleviation on the archaeological record of the central and eastern Great Plains, U.S.A. Geoarchaeology: an International Journal 17(2), 141- 154. Bonhomme, T. (1983). Bunda Lake – a Study in Site Location. Unpublished B.A. (Hons.) thesis, Macquarie University. Bowler, J.M. (1970). Late Quaternary Environments: A Study of Lakes and Associated Sediments in South Eastern Australia. Unpublished PhD Thesis, Australian National University, Canberra. Bowler, J.M. (1971). Pleistocene salinities and climate change: evidence from lakes and lunettes in southeastern Australia. In D.J. Mulavaney & J. Golson (eds) Aboriginal Man and Environment in Australia, pp. 47–65. Australian National University Press: Canberra. Bowler, J.M. (1973). Clay dunes: their occurrence, formation and environmental significance. Earth Science Reviews 9, 315-318.

19 Chapter 1: Introduction Bowler, J.M. (1976). Recent developments in reconstructing late Quaternary environments in Australia. In R.L. Kirk & A.G. Thorne (eds) The Origin of the Australians, pp. 55-77. Humanities Press: New Jersey. Bowler, J.M., Jones, R., Allen, H.R. & Thorne, A.G. (1970). Pleistocene human remains from Australia: a living site and human cremation from Lake Mungo. World Archaeology 2, 39-60. Bowler, J.M., Thorne, A.G. & Polach, H.A. (1972). Pleistocene man in Australia: age and significance of the Mungo skeleton. Nature 240, 48-50. Bowler, J.M., Hope, G., Jennings, J., Singh, G. & Walker, D. (1976). Late Quaternary climates of Australia and New Guinea. Quaternary Research 6, 359-394. Buck, B.J., Steiner, R.L., Burgett, G. & Monger, H.C. (1999). Artefact distribution and its relationship to microtopographic geomorphic features in an eolian environment, Chihuahuan Desert. Geoarchaeology: an International Journal 14(8), 735-754. Butzer, K.W. (1982). Archaeology as Human Ecology: Method and Theory for a Contextual Approach. Cambridge University Press: Cambridge. Chorley, R.J., Schumm, S.A. & Sugden, D.E. (1984). Geomorphology. Methuen: London. David, B., Roberts, R., Tuniz, C., Jones, R. & Head, L. (1997). New optical and radiocarbon dates from Ngarrabullgan Cave, a Pleistocene archaeological site in Australia: implications for the comparability of time clocks and for the human colonization of Australia. Antiquity 71, 183-188. Davidson, D.A. & Shackley M.L. 1976. Geoarchaeology: earth science and the past. Duckworth: London. Doleman, W.H. (1992). Geomorphic and behavioral processes affecting site and assemblage formation: the EARM data. In W.H. Doleman, R.C. Chapman, R.L. Stauber & J. Piper (eds) Landscape Archeology in the Southern Tularosa Basin. Vol. 3. Archeological Distributions and Prehistoric Human Ecology. OCA/UNM Report No. 185-324F, pp. 73-105. Office of Contract Archeology, University of New Mexico: Albuquerque. Doleman, W.H. & Stauber, R.L. (1992). Macrotopographic landform analysis as a predictor of archeological remains on the Tularosa Basin floor. In W.H. Doleman, R.C. Chapman, R.L. Stauber & J. Piper (eds) Landscape Archeology in the Southern Tularosa Basin. Vol. 3. Archeological Distributions and Prehistoric Human Ecology. OCA/UNM Report No. 185-324F, pp. 107-151. Office of Contract Archeology, University of New Mexico: Albuquerque.

20 Chapter 1: Introduction Doleman, W.H., Chapman, R.C., Stauber, R.L. & Piper J. (1992). Landscape Archeology in the Southern Tularosa Basin. Vol. 3. Archeological Distributions and Prehistoric Human Ecology. OCA/UNM Report No. 185-324F. Office of Contract Archeology, University of New Mexico: Albuquerque. Dortch, C. (1979). Devil’s Lair, an example of prolonged cave use in southwestern Australia. World Archaeology 10, 258-279. FANNING, P.C. (1999). Recent landscape history in arid western NSW, Australia: a model for regional change. Geomorphology 29, 191 – 209. Fankhauser, B. L. & Bird, J. R. 1993 Archaeometry: current Australasian research. Occasional Papers in Prehistory No. 22, Department of Prehistory, Research School of Pacific Studies, Australian National University, Canberra. Friedel, M.H., Foran, B.D. & Stafford Smith, D.M. (1990). Where the creeks run dry or ten feet high: pastoral management in arid Australia. Proceedings of the Ecological Society of Australia 16, 185–194. Fullagar, R.L.K., Price, D.M. & Head, L.M. (1996). Early human occupation of Northern Australia: archaeology and thermoluminescence dating of Jimnium rock-shelter, Northern Territory. Antiquity 70, 751-773. Gladfelter, B. G. (1977). Geoarchaeology: the geomorphologist and archaeology. American Antiquity 42(4), 519-538. Gladfelter, B.G. (1981). Developments and directions in geoarchaeology. Advances in Archaeological Method and Theory 4, 343-364. Goldberg, P., Holliday, V.T. & Ferring, C.R. (2001). Earth Sciences and Archaeology. Kluwer: New York. Hale, H. & Tindale, N. (1930). Notes on some human remains in the lower Murray Valley, South Australia. Records of the South Australian Museum 4, 145–218. Hallam, S. J. (1977). Topographic archaeology and artefactual evidence. In R. V. S. Wright (ed.) Stone Tools as Cultural Markers: Change, Evolution and Complexity, pp. 169– 177. Australian Institute of Aboriginal Studies: Canberra. Harris, J.W.K. (1980). Early man. In A. Sherratt (ed.) The Cambridge Encyclopedia of Archaeology, pp. 62 – 70. Cambridge University Press: London. Herz N. & Garrison E.G. (1998). Geological Methods for Archaeology. Oxford University Press: New York. Hiscock, P. & Hughes, P.J. (1983). One method of recording scatters of stone artefacts during site surveys. Australian Archaeology 17, 87-98.

21 Chapter 1: Introduction Holdaway, S.J. & Stern, N. (in press). Written in Stone: Methods and Theories for Analysing Australian Flaked Stone Artefacts. Scholastic Press: Melbourne. HOLDAWAY, S., FANNING, P. & WITTER, D. (1997). GIS analysis of artefact distributions in an eroding landscape: the Western New South Wales Archaeological Project. In Johnson, I. & North, M. (eds) Archaeological Applications of GIS: Proceedings of Colloquium II, UISPP XIIIth Congress, Forli, Italy, September 1996. Sydney University Archaeological Methods Series 5. HOLDAWAY, S., WITTER, D., FANNING, P., MUSGRAVE, R., COCHRANE, G., DOLEMAN, T., GREENWOOD, S., PIGDON, D. & REEVES, J. (1998). New approaches to open site spatial archaeology in Sturt National Park, New South Wales, Australia. Archaeology in Oceania 33, 1-19. Hughes, P.J. & Sullivan, M.E. (1981). Aboriginal burning and late Holocene geomorphic events in eastern New South Wales. Archaeology and Physical Anthropology in Oceania 8, 277-278. Hughes, P.J. & Sullivan, M.E. (1982). Geoarchaeology in Australia: a review. In W. Ambrose & P. Duerden (eds) Archaeometry: an Australasian perspective, pp. 100- 111. Australian National University Press: Canberra. Jones, R. (1968). The geographical background to the arrival of man in Australia and Tasmania. Archaeology and Physical Anthropology in Oceania 3, 186-215. Jones, R. (1969). Fire-stick farming. Australian Natural History 16, 224-228. Kuehn, D.D. (1993). Landforms and archaeological site location in the Little Missouri badlands: a new look at some well-established patterns. Geoarchaeology: an International Journal 8(4), 313-332. McBryde, I. (1968). Archaeological investigations in the Graman district. Archaeology and Physical Anthropology in Oceania 3, 77–93. McBryde, I. (1974). Aboriginal Prehistory in New England. Sydney University Press: Sydney. Mulvaney, D.J. (1969). The Prehistory of Australia. Thames and Hudson: London. Mulvaney, D.J. & Joyce, E.B. (1965). Archaeological and geomorphological investigations on Mt. Moffatt Station, Queensland, Australia. Proceedings of the Prehistoric Society 31, 147-212. Nash, D.T. & Petraglia, M.D. (1987). Natural Formation Processes and the Archaeological Record. BAR International Series 352. British Archaeological Reports: Oxford. O’Connell, J. & Allen, J. (1998). When did humans first arrive in Greater Australia and why is it important to know? Evolutionary Anthropology 6(4), 132-146.

22 Chapter 1: Introduction O’Connor, S. (1995). Carpenter’s Gap Rockshelter 1: 40,000 years of Aboriginal occupation in the Napier Ranges, Kimberley, WA. Australian Archaeology 40, 58-59. Petraglia, M. D. & Nash, D. T. (1987). The impact of fluvial processes on experimental sites. In D. T. Nash & M. D. Petraglia (eds) Natural Formation Processes and the Archaeological Record, pp. 108-130. BAR International Series 352. British Archaeological Reports: Oxford. Pollard, A. M. (1999, ed.). Geoarchaeology: Exploration, Environments, Resources. Geological Society, London, Special Publications 165. Prescott, J.R. (1988). Archaeometry: Australasian Studies 1988. Department of Physics and Mathematical Physics, the University of Adelaide: Adelaide. Rapp, G. Jr. & Gifford, J. (1985). Archaeological Geology. Yale University Press: New Haven. Rapp, G. Jr. & Hill, C. (1998). Geoarchaeology. Yale University Press: New Haven. Rick, J. W. (1976). Downslope movement and archaeological intrasite spatial analysis. American Antiquity 41, 133-144. Roberts, R.G., Jones, R. & Smith, M.A. (1990). Thermoluminescence dating of a 50,000- year-old human occupation site in northern Australia. Nature 345, 153-156. Robins R.P. (1993). Archaeology and the Currawinya Lakes: Towards a Prehistory of Arid Lands of Southwest Queensland. Unpublished PhD thesis, Griffith University, Brisbane. Robins R.P. (1996). A report on archaeological investigations of open hearth sites in southwest Queensland. Queensland Archaeological Research 10, 25-35 Robins R.P. (1997). Patterns in the landscape: a case study in nonsite archaeology from southwest Queensland. Memoirs of the Queensland Museum, Cultural Heritage Series 1(1), 23-56. Robins R.P. (1998). Archaeological investigations at Youlain Springs, southwest Queensland. Memoirs of the Queensland Museum, Cultural Heritage Series 1(1), 57- 74. Robins, R. (1999). Lessons learnt from a taphonomic study of stone artefact movement in an arid environment. In M.J. Mountain & D. Bowdery (eds) Taphonomy: the Analysis of Processes from Phytoliths to Megafauna. Department of Archaeology and Natural History, Research School of Pacific Studies, Research Papers in Archaeology and Natural History No. 30, pp. 93-106. Australian National University: Canberra. Ross, A. (1981). Holocene environments and prehistoric site patterning in the Victorian Mallee. Archaeology in Oceania 16, 145-154.

23 Chapter 1: Introduction Ross, A. (1984). If There Were Water: Prehistoric Settlement Patterns in the Victorian Mallee. Unpublished PhD thesis, Macquarie University, Sydney. Ross, A., Donnelly, T. & Wasson, R.J. (1992). Peopling of the arid zone: human- environment interactions. In J.R. Dodson (ed.) The Naive Lands: Prehistory and Environmental Change in Australia and the Southwest Pacific, pp. 76-114. Longman Cheshire: Melbourne. Rossignol, J. (1992). Concepts, methods, and theory building. In J. Rossignol, & L.A. Wandsnider (eds) Space, Time and Archaeological Landscapes, pp. 3-16. Plenum: New York. Rossignol, J. & Wandsnider, L.A. (1992). Space, Time and Archaeological Landscapes. Plenum: New York. Schick, K. D. (1987). Experimentally-derived criteria for assessing hydrologic disturbance of archaeological sites. In D. T. Nash & M. D. Petraglia (eds.) Natural Formation Processes and the Archaeological Record, pp. 86-107. BAR International Series 352. British Archaeological Reports: Oxford. Schiffer, M.B. (1972). Archaeological context and systemic context. American Antiquity 37, 156-165. Schiffer, M.B. (1983). Towards identification of site formation processes. American Antiquity 48, 675-706. Seaman, T.J., Doleman, W.H. & Chapman, R.C. (1988) The Border Star 85 Survey: Toward an Archeology of Landscapes. University of New Mexico Office of Contract Archaeology Project No. 185-227. University of New Mexico: Albuquerque. Shackley, M. L. (1978). The behaviour of artefacts as sedimentary particles in a fluviatile environment. Archaeometry 20, 55-61. Shawcross, F.W. & Kaye, M. (1980). Australian archaeology: implications of current interdisciplinary research. Interdisciplinary Science Reviews 5(2), 112-127. Smith, M.A. (1987). Pleistocene occupation in arid Central Australia. Nature 328, 710-711. Smith, M.A. (1988). The Pattern and Timing of Prehistoric Settlement in Central Australia. Unpublished PhD thesis, University of New England, Armidale. Smith, M.A. (1989). The case for a resident human population in the Central Australian ranges during full glacial aridity. Archaeology in Oceania 24, 93-105. Stein, J.K. (1983). Earthworm activity: a source of potential disturbance of archaeological sediments. American Antiquity 48, 277-289.

24 Chapter 1: Introduction Stein, J.K. (2001). A review of site formation processes and their relevance to geoarchaeology. In P.Goldberg, V.T. Holliday & C.R. Ferring (eds) Earth Sciences and Archaeology, pp. 37-51. Kluwer: New York. Thomas, D.H. (1975). Nonsite sampling in archaeology: up the creek without a site? In J.W. Mueller (ed.) Sampling in Archaeology, pp. 61-81. University of Arizona Press: Tuscon. Thorley P.B. (1998a). Shifting Location, Shifting Scale: a Regional Landscape Approach to the Prehistoric Archaeology of the Palmer River Catchment, Central Australia. Unpublished PhD thesis, Northern Territory University, Darwin. Thorley P.B. (1998b). Pleistocene settlement in the Australian arid zone: occupation of an inland riverine landscape in the central Australian ranges. Antiquity 72, 34-45. Thorley P.B. (1999). Regional archaeological research in the Palmer River catchment. Australian Aboriginal Studies 1999(2), 62-68. Thorley P.B. (2001). Uncertain supplies: water availability and regional archaeological structure in the Palmer River catchment, central Australia. Archaeology in Oceania 36, 1-14. Thorne, A.G., Grun, R., Mortimer, G., Spooner, N.A., Simpson, J.J., McCulloch, M.M., Taylor, L. & Curnoe, D. (1999). Australia’s earliest human remains: age of the Lake Mungo 3 skeleton. Journal of Human Evolution 36, 591-612. Turney, C.S.M., Bird, M.I., Fifield, L.K., Roberts, R.G., Smith, M.A., Dortch, C.E., Grun, R., Lawson, E., Ayliffe, L.A., Miller, G.H., Dortch, J. & Cresswell, R.G. (2001). Early human occupation at Devil’s Lair, southwestern Australia 50,000 years ago. Quaternary Research 55, 3-13. Veth, P. (1993). Islands in the Interior: the Dynamics of Prehistoric Adapations within the Arid Zone of Australia. Archaeological Series 3, International Monographs in Prehistory: Ann Arbor. Villa, P. (1982). Conjoinable pieces and site formation processes. American Antiquity 47, 276-290. Villa, P. & Courtin, J. (1983). The interpretation of stratified sites: a view from the underground. Journal of Archaeological Science 10, 267-281. Wandsnider, L.A. (1989). Long-Term Land Use, Formation Processes, and the Structure of the Archaeological Landscape: a Case Study from Southwestern Wyoming. Unpublished PhD thesis, University of New Mexico, Albuquerque.

25 Chapter 1: Introduction Wandsnider, L.A. (1992). Archaeological landscape studies. In J. Rossignol, & L.A. Wandsnider (eds) Space, Time and Archaeological Landscapes, pp. 285-292. Plenum: New York. Waters, M.R. (1991). The geoarchaeology of gullies and arroyos in southern Arizona. Journal of Field Archaeology 18(2), 141-159. Waters, M.R. (1992). Principles of Geoarchaeology: a North American Perspective. University of Arizona Press: Tuscon. Waters, M.R. (2000). Alluvial stratigraphy and geoarchaeology in the American southwest. Geoarchaeology: an International Journal 15(6), 537-557. Waters, M.R. & Kuehn, D.D. (1996). The geoarchaeology of place: the effect of geological processes on the preservation and interpretation of the archaeological record. American Antiquity 61(3), 483-497. Wells, L.E. (2001). A geomorphological approach to reconstructing archaeological settlement patterns based on surficial artefact distribution. In P. Goldberg, V.T. Holliday & C.R. Ferring (eds) Earth Sciences and Archaeology, pp.107-141. Kluwer: New York. Williams, M.A.J. (1982). Alluvial stratigraphy and archaeology: three examples from prehistoric Africa, India and Australia. In W. Ambrose & P. Duerden (eds) Archaeometry: an Australasian Perspective, pp. 112-119. Australian National University Press: Canberra. Witter, D.C. (1992). Regions and Resources. Unpublished PhD thesis, Australian National University, Canberra. Wood, W. R. & Johnson, D. L. (1978). A survey of disturbance processes in archaeological site formation. Advances in Archaeological Method and Theory 1, 315-381. Worrall, L. (1980). Geoarchaeological Investigations at Mangrove Creek N.S.W. Unpublished B.A. (Hons.) thesis, Australian National University, Canberra. Zvelebil, M., Green, S.W. & Macklin, M.G. (1992). Archaeological landscapes, lithic scatters and human behavior. In J. Rossignol, & L.A. Wandsnider (eds) Space, Time and Archaeological Landscapes, pp. 193-226. Plenum: New York.

26 Archaeol. Oceania 33 (1998) 1-19

New approaches to open site spatial archaeology in Sturt National Park, New South Wales, Australia

SIMON HOLDAWAY, DAN WITTER, PATRICIA FANNING, ROBERT MUSGRAVE, GRANT COCHRANE, TRUDY DOELMAN, SIMON GREENWOOD, DAN PIGDON and JAMIE REEVES

Abstract

Surface scatters of stone artefacts are ubiquitous in the Australian ological record. Rather than look at the distribution of landscape and form the basis for the majority of archaeological sites across a region, we argue that the conventional con­ conservation decisions. The research reported here proposes a dis­ cept of a site is unproductive and that regional studies tributional approach for analysing this record founded on the arte­ should begin by developing an understanding of the way fact as the minimal recording unit rather than the site. A method of the archaeological record has formed on an artefact by assemblage definition is proposed to permit the study of assem­ artefact basis. Artefacts may then be combined into spa­ blage composition across space. The method is applied to artefacts tially defined assemblages and variability in assemblage exposed on the surface as a result of recent erosion at TIB 13 (Sturt composition investigated at a landscape level. We report National Park, NSW). All stone artefacts greater than 20mm in here the results of one such analysis of TIB 13, a location maximum dimension were recorded by locating each artefact in in Sturt National Park, NSW (Witter 1992). three-dimensional space and analysing it in place. The distribution of these artefacts was then compared to the nature of the landform on which they rested through the use of a GIS. Analysis of assem­ The definition of stone artefact assemblages blage composition indicates significant differences across an area of approximately 30,000m2. Whether archaeologists excavate artefacts from stratified deposits in rockshelters or collect those distributed on the surface, they analyse artefacts by using their common Surface scatters of stone artefacts are the most common location in time and space to search for pattern represent­ phenomenon in the archaeology of Aboriginal sites in ing repeated behaviour. Only rarely can we see the indi­ Australia yet their study has probably contributed the vidual at work in such situations; instead an average pic­ least to our understanding of Aboriginal prehistory. The ture of behaviour is reconstructed from the material reasons why are not difficult to find. Open sites normally abandoned by many individuals at one particular loca­ lack the stratigraphy that is fundamental to the analysis of tion. For stone artefacts this behavioural average is most many sets of artefacts, they are areally extensive with no clearly seen by studying processes such as procurement, clear boundaries, they contain few features with which to manufacture, use and discard. demarcate groups of stone artefacts, the stone artefacts In rockshelters the walls ultimately limit space, themselves are not easy to interpret, and their identifica­ although more often than not the limits are those of the tion is controlled by exposure and visibility related to excavation unit; time is proscribed using a number of ground surface conditions. Yet because they are ubiqui­ techniques (in Australia those that provide absolute age tous, open site stone artefact scatters are the bread and determinations are the most common). The assemblages butter of much field archaeology in Australia, particularly of artefacts created by the use of time and space units that connected with Environmental Impact Assessment reflect a particular scale of resolution. Even the best- (EIA), and they are the subject of many of the major dated site (where chronology relies on radiocarbon) will archaeological conservation decisions made in Australia. have a temporal resolution measured in decades or cen­ This paper outlines initial results of a project aimed at turies. This means that archaeologists, of necessity, developing new ways of dealing with this form of archae- analyse long term accumulations of artefacts no matter what type of record they investigate. In this sense studying a surface scatter is no different SH, GC, TD, SG, JR: Department of Archaeology, La Trobe to excavating in rockshelters. In both situations artefacts University, Bundoora, Vic 3083; accumulate through time and are grouped together on DW: NSW National Parks and Wildlife Service, Broken Hill, the basis of their common location. Where these types NSW; of record differ is in the way time may be controlled: PF: Graduate School of the Environment, Macquarie University, North Ryde, NSW; stratified rockshelters offer the opportunity to study RM: School of Earth Sciences, La Trobe University; artefact accumulation through time via the creation of DP: Department of Geomatics, University of Melbourne, multiple temporal units whereas surface scatters nor­ Parkville, Vic; mally offer the opportunity to define only a single Ms submitted 6/97, accepted 10/97 chronological unit. Rockshelters and surface scatters

27 Chapter 1: Introduction Holdawayetal.(1998)

differ, therefore, in terms of our ability to categorise Criticisms of the site concept are not new (see time, but not necessarily in the level of precision with Dunnell & Dancey 1983), and overseas at least, much which such units are defined. If the age of the surface on work has been devoted to developing alternatives to the which artefacts lie can be determined this in effect dates site based approach culminating in a variety of distribu­ the artefacts at a precision given by this surface's age. tional and landscape archaeologies (Ebert 1992; While many surfaces may accumulate artefacts over Rossignol & Wandsnider 1992). In many such studies millennia rather than decades or centuries, a chronology 'site' has been replaced by 'artefact' as the minimal with a precision measured in millennia is not unknown recording unit. But while using the artefact as the record­ for rockshelter sites (Cosgrove 1995). Of course, what is ing unit gets around the problem of site boundaries, it lost in the ability to study chronological change through introduces a different problem, that of defining assem­ time with surface scatters is replaced by the ability to blages. Since the site forms the ultimate unit from which investigate change across space. While rockshelter sites an assemblage may be defined, removing this assem­ sometimes show intrasite differences there is a much blage definition is problematic. greater opportunity to investigate spatial variability in Those interested in landscape approaches have turned surface scatters where the effect of natural boundaries is to techniques developed for intrasite analysis to solve much reduced. this problem (Ebert 1992). Early intrasite studies relied This study aims to investigate stone artefact assem­ heavily on direct ethnographic analogy with assemblages blage variability by analysing surface scatters, rather than defined in terms of "living floors', "activity areas', and stratified rockshelter deposits, to determine whether vari­ "toolkits' (Freeman & Butzer 1966; Whallon 1973, ability in the spatial deposition of artefacts in the past has 1974). But this has changed in the face of criticisms that left patterns that can be detected archaeologically. The co-variation in space need not necessarily reflect a sim­ example we discuss draws on work being conducted in ple functional association (Schiffer 1972:161-2) and eth- Sturt National Park under the auspices of the Western noarchaeological evidence that activity areas may not be New South Wales Archaeological Program (WNSWAP). discernible in the archaeological record (O'Connell et al. One aspect of this program involved the location and 1991:73-4; Yellen 1977:85-6). With a few exceptions analysis of 10,000 stone artefacts across an area of 29,930 (Carr 1984:106), these criticisms have been accepted; m2 identified as TIB 13 (Witter 1992:142). All data researchers are no longer searching for short term ethno­ retrieved has been integrated into geographic information graphic explanations for intra-site patterns in assemblage system (GIS) software (ARCINFO and ARCVIEW). In composition. this paper we discuss the rationale for using this software, The change in theoretical orientation should not, how­ the field methods adopted, and the analytical techniques ever, be seen as diminishing the utility of the methods developed using the GIS. We provide details of the analy­ developed for intrasite analysis. Many of the techniques sis of this data to illustrate the type of results we hope to are useful for assemblage based pattern recognition; it is recover during future periods of research. the explanation of the pattern that has changed. Rather than the identification of tool kits and activity areas, eth- noarchaeological research is increasingly being directed Alternatives to sites as an analytical category at asking how and why behaviour is organised as it is at particular locations, how this organisation is reflected in Without rockshelters, the boundaries of a site are much the distribution of refuse and whether knowledge of harder to determine; in fact, in the absence of clear evi­ these relationships can be applied to the types of pattern dence that occupation was spatially constrained (perma­ apparent in time transgressive archaeological contexts nent houses for instance), the use of a site as an analyti­ (O'Connell era/. 1991). cal unit becomes problematic (Ebert 1992; Dunnell Recent ethnoarchaeological work provides a number 1992). This is particularly true in Australian archaeology of observations relevant to the interpretation of surface where surface scatters of artefacts predominate. In her­ stone artefact scatters. It tells us, for instance, that the itage management, for instance, where the concept of a longer the duration of occupation, the greater the 'site' is central to the entire industry (and to conservation chance that there will be patterns in the distribution of decision making) definitional problems abound. The abandoned artefacts. We may expect a more structured New South Wales National Parks and Wildlife Service distribution of artefacts near resources when their util­ originally considered that two artefacts within 50m of ity, nature and location, required extended time for each other constituted a site. This is no longer considered exploitation (O'Connell 1987; O'Connell et al. 1991; appropriate (NSW NPWS 1997) but it has proved diffi­ Cameron & Tomka 1993; Wandsnider 1996). Patterns cult to provide a definition for documenting the distribu­ in the distribution of artefacts will be apparent over tion of archaeological evidence in units suitable for man­ large areas (thousands of square metres) and identifi­ agement purposes. Clearly, different units need to be able as general trends in the frequency and size of considered when describing the archaeological record for abandoned artefacts. And they will probably not be vis­ management purposes and for assessment in terms of ible with the conventional archaeological samples of a prehistory, but few techniques for doing this have so far few tens of square meters or less (Wobst 1983). These been proposed. studies suggest that we should not be attempting to

28 Holdawayetal.(1998) Chapter 1: Introduction

Bulloo Embayment

Figure 1. Location Map.

identify living floors, activity areas or toolkits but The study area — TIB 13 developing methodologies capable of defining assem­ blages without recourse to sites across areas that are Since WNSWAP was conceived to promote new meth­ very large compared to those recorded in conventional ods for tackling the archaeology of surface stone artefact archaeological fieldwork. scatters, the project personnel were recruited from a vari­ It is our suggestion that we can undertake such stud­ ety of backgrounds to ensure a multi-disciplinary per­ ies by adapting the techniques applied to intrasite stud­ spective. We currendy have a geomorphologist (PF), a ies, using them to discover patterns in surface artefact heritage manager (DW) and an academic archaeologist scatters. What is needed is a set of methods for defin­ (SH) together with a geophysicist (RM) working on the ing assemblages (one of which we describe below) and project. As a consequence, the project has developed a a set of techniques for analysing the composition of the number of research areas designed to address the four resulting assemblages in terms of artefact manufacture. problems with surface scatter archaeology identified Here, we provide one set of examples based on our above: work at TIB 13 where we seek patterns in assemblage • absence of clear boundaries, composition across space that can be used to define long term trends (just how long, and how time is to be • the lack of methods for grouping artefacts into assem­ defined, is discussed below) in the way refuse was dis­ blages for analysis, carded across a landscape in the past. Given sufficient • the difficulties involved in extracting information research, these patterns may be able to be interpreted from stone artefacts, and in terms of stone resource exploitation, manufacture, use and discard within a particular geographic region • the problem of chronology when faced with the lack but at present we limit our discussion to the techniques of stratigraphy. that enable us to identify these patterns at a single The location we discuss, TIB 13, was originally iden­ location. tified by one of us (DW) during an earlier phase of

29 Chapter 1: Introduction Holdaway et al. (1998)

research at Sturt National Park. It was selected as a loca­ Field methods tion to trial the recording and analytical systems we describe here because it seemed to encompass a broad The field methods we developed for work at TIB 13 range of features typical of archaeological deposits in the reflect both the erosional characteristics of the region region within one relatively constrained area. Of these, and our interest in recording individual artefacts as the the most critical was the degree of erosion and stripping minimal depositional and analytical units. Because we of the landsurface which has occurred since sheep graz­ are dealing with such a large area and many artefacts, we ing was introduced to the region in the late 1800s. In considered the use of a graphical relational database (a effect this erosion has 'excavated' an archaeological vector Geographic Information System (GIS) — 'site' for us, exposing many thousands of artefacts on ARCINFO) essential to our operation. lagged surfaces. For archaeologists interested in using The field methods consisted of two parts: the artefact as the minimal depositional and analytical unit, this erosion has produced the ideal landscape. We • the construction of a map of the geomorphological have tens of thousands of squares meters of 'excavated' features within the study area, and land on which to search for interpretable patterns of arte­ • recording the location and nature of the stone artefacts fact deposition. Moreover, we are able to do this without distributed across the landscape. further disturbance to the material, acceding to the Experience showed that these two tasks were con­ wishes of the Wangkumara people, on whose land we are ducted most efficiently by two separate crews. Ideally, working, that their heritage be left in place. geomorphological mapping of a segment of the study The TIB 13 location is adjacent to an unnamed left area should be completed before the artefact-recording bank tributary of Thomsons Creek within a kilometre of crews proceeded with their tasks. the Mt. Wood homestead in Sturt National Park (Figure All mapping was conducted in three-dimensional 1). A low escarpment comprising Cretaceous marine Australian Map Grid (AMG) coordinates from a datum sediments capped with silcrete forms the eastern bound­ location established with a Global Positioning System ary, with the creek forming the western boundary. In (GPS). This was done to allow seamless integration with between is a flat valley floor surface underlain by allu­ other data mapped in AMG (i.e. commercially available vium. It is currently devoid of vegetation, except for a digital maps). It is important to note that AMG coordi­ few gidgee {Acacia cambagei) along the watercourse and nates are double precision numbers. This can cause prob­ scattered chenopod shrubs and grasses, as well as mulga lems for some software packages. (Acacia aneura), on the slope below the escarpment. Rainwash and wind erosion has removed topsoil down to the level of the hardsetting bleached A2 horizon of the Geomorphological mapping original duplex soil profile, which has formed in the allu­ vium. Incision below this level has occurred along rills, Geomorphological mapping aimed to record the nature exposing the domed columnar structure of the red silty of the land-surfaces on which the artefacts were resting clay subsoil. Gravel is scattered over much of the valley in terms of their depositional or erosional history. A floor surface as a lag, except where it has been buried by modified version of the regolith/terrain mapping system transported sediment from upslope, forming features of Pain et al. (1991) was used. Four landform elements which have been termed 'sediment islands'. were mapped (Figure 2): erosional surfaces that retained Along the watercourse, deposits consisting of sands, gravel and artefacts as a lag; areas of sediment deposi­ granules and gravel in a grey clay matrix, which proba­ tion; incised depressions into the landsurface such as bly accumulated in ephemeral waterholes, have been rills, gullies and channels; and the strip of land adjacent overlain by at least 30cm of red sandy sediments derived to the main stream channel where overbank deposition of from the topsoils eroded from catchment slopes. This alluvium was dominant and almost completely masked material is variously referred to as either "post-European the archaeological record. As will be detailed below, material' or 'post-settlement alluvium' (PSA). these mapped surfaces were then analysed in terms of Radiocarbon determinations of charcoal from Aboriginal their effect on artefact visibility and taphonomy. fireplaces buried by this alluvium at Mootwingee and Mapping was achieved using an electronic total station Fowlers Gap (Fanning 1996) indicate that it postdates (Sokkia SET SC) connected to a hand held computer European occupation of the region and it is therefore (SDR 33), using a crew of three people. A geomorpholo- likely to be the product of the initial phase of disturbance gist (PF) determined the landform element boundaries of the vegetation cover by sheep grazing and associated and recorded the details in a field notebook. activities. Channels subsequently incised the valley floors, exposing these materials in the channel banks. The channels are continuing to enlarge by incision, Artefact mapping widening and knick-point retreat since the hydrodynamic instability initiated by land-cover change in the late nine­ Details of artefacts were recorded in two separate data­ teenth century has not completely worked its way bases that were linked by the use of a common identifi­ through the upland catchment systems (Fanning 1996). cation number for each artefact: a locational database to

30 Holdawayetal.(1998) Chapterl: Introduction

Channel 2 Channel 1

Artefact Depositional Lag Rills, gullies, channels Stream margin

100m

Figure 2. Landsurface types and artefact distribution at TIB 13.

31 Chapter 1: Introduction Holdaway et al. (1998)

Attribute Value Description

Data-class Complete flake Has a platform and a termination Proximal flake Retains a platform and no termination Distal flake A termination with no platform Medial flake No platform and no termination Complete tool, proximal tool, As above, but with macroscopic retouch. distal tool, medial tool. Core Negative flake scars including both producer cores and nuclear tools

Material Quartzite Silcrete Quartz

Distal end Feather Tapering termination Abrupt Non-tapering termination Plunge Curves toward the ventral surface Hinge Curves toward the dorsal surface

Form Blade Parallel flake scars Expanding Proximal end narrower than distal end Intermediate All other flake forms n/a Form cannot be determined

Platform type Unifacial Struck from a unifacially flaked or cortical platform Bifacial Struck from a bifacially flaked platform

Tool type Scraper Continuous macroscopic scalar or stepped retouch Notch Retouch forming one or more single cuspate notches Utilised Edge nibbling that may be discontinuous Point/backed blade Backed blades and unifacially retouched points Adze Tula or burren

Core types Unifacial Platforms flaked from a single direction Bifacial Platforms flaked from two directions Microblade Multiple parallel flake scars across core surface

Dimensions Maximum length Longest dimension in any axis. Measured on all pieces Maximum width At right angles to maximum length, only on complete flakes and tools Maximum thickness Where length and width intersect, only on complete flakes and tools

Table 1. Artefact attributes and values recorded at TIB 13.

32 Holdawayetal.(1998) Chapter!: Introduction

Land-surface Number of Area in m2 Artefacts Results artefacts per m1 Analysis using a GIS 608 4475.2 0.14 567 5969.5 0.09 The GIS allows us to integrate the locational and attribute databases for the artefacts together with the 7856 13972.8 0.56 locational database for the geomorphological land-sur­ 634 5132.9 0.12 faces. Database queries may be constructed that com­ 169 379.9 0.44 bine any of the spatial and analytical data that we recorded. The practical application of this software is Table 2. Landsurface areas and artefact density. illustrated in a series of analyses below. We begin by considering the effect of the erosional processes that have exposed the artefacts, particularly issues of arte­ fact visibility and movement by water. We then intro­ record the three dimensional spatial co-ordinates of the duce a method for assemblage definition based on the artefact, and an attribute database that contained values spatial distribution of artefacts. The results of analyses for a range of technological and typological variables. based on this method are presented to show the type of Once the geomorphological mapping of a part of the assemblage level spatial patterning present at TIB 13. study area was completed, a large crew of archaeological We finish with a brief discussion of our future analyti­ field workers surveyed die area. To ensure that the total cal goals. area was covered, this survey was very intensive, with workers separated by only a metre walking in several lines abreast. All artefacts with a maximum dimension of Artefact density on land-surfaces 20mm or more (after Schick 1987 — see below) were identified. When an artefact was found, a nail with a Figure 2 illustrates the pattern of artefact dispersal across coloured tape attached was placed next to it The survey four landform elements identified at TEB 13. It is clear crew of three then mapped each artefact. The coordinates from this figure that artefact density varies considerably were stored in the SDR memory, and a sequential ID with landsurface type, and this is confirmed when true number allocated. The EDM operator advised the target densities are calculated (see Table 2). Area can be easily holder of the number via radio contact, and this was writ­ calculated for each landform element with the GIS soft­ ten on the coloured tape. ware, as can the number of artefacts that are located on Once the location of the artefacts had been recorded, a each element. separate team recorded technological and typological The results indicate two sets of density values. information for each artefact. The list of attributes Depositional and Stream Margin landforms have low recorded and their definitions is provided in Table 1. densities, around 0.1 artefacts per square meter, while Because TIB 13 was recorded at an early stage of the Lag and Rill landforms have densities that are four to project a very basic set of attributes were used. five times higher, around 0.5 artefacts per square metre. Subsequent work at Sturt has expanded the range of It is likely that this difference in density reflects visibility attributes taken. differences; some artefacts in the Depositional and Recording was achieved by logging directly into Stream Margin landforms are buried so were not palmtop computers (HP 200LX) running data entry soft­ recorded (cf. Witter 1992:84-7). Alternatively, there may ware (ENTRER TROIS [McPherron & Holdaway have been fewer artefacts deposited in the Depositional 1996]). This software prompts the users for input making Surface and Stream Margin areas in prehistory. Channels extensive use of menus for nominal or ordinal values. have the lowest density of artefacts reflecting removal by The identification number on the coloured tape was water flow (see below). recorded for each artefact uniquely identifying it and One way to differentiate between these possibilities providing a link to the locational database. would be to excavate the Depositional Surface and After each day's work, the co-ordinate data and arte­ Stream Margin landforms and record the distribution of fact descriptions were downloaded from the data loggers buried artefacts in each. While possible, such an under­ and palmtops. Co-ordinates were stored both in a con­ taking would be extremely time consuming, and as dis­ ventional relational database and as converted GIS ele­ cussed above, not likely to win support from either the ments (points, lines and polygons). For our purposes Wangkumara Aboriginal Community or the NSW land-surfaces were recorded as polygons (a sequence of NPWS. The alternative, which is discussed below, is to points joined by lines that close back to the starting develop techniques mat enable us to exclude landforms point) and sometimes lines (a series of points connected that may have lower densities of artefacts, possibly as a by lines that do not close). Artefact locations were result of visibility problems, without compromising the recorded as points. The GIS elements were then labelled ability to analyse the spatial distribution of artefacts according to the type of land-surface or artefact form across the site as a whole. they represented (Figure 2). In the analyses that follow we treat TIB 13 as an ana-

33 Chapter 1: Introduction Holdaway et al. (1998)

lytical location by looking for patterns in the distribution Number of Mean maximum Standard of artefacts on the Lag Surface. In this case the location Artefacts dimension deviation is bounded by Depositional Surface and Stream Margin landform elements that totally surround the Lag Surface Channel 1 297 37.1 14.5 element. The location boundary is definable but is still arbitrary in the sense that the land-surfaces that define it 0-2m buffer 269 34.6 13.0 are the result of recent geomorphological changes 2-4m buffer 249 33.8 11.5 removed in time from the date when the artefacts were 4-6m buffer 194 35.6 13.3 deposited. The simple analysis presented here demonstrates the Artefact mean maximum dimension in Channel 1 and 2, point made some years ago (Witter 1992) that surface 4 and 6m buffers from channel. F = 3.21, d/= 3, 1005, exposures are liable to be quite variable depending on p = 0.02 the nature of the surface geomorphology. Simply attempting to assess artefact densities (a measure that has in the past been used as an indicator of site significance Number of Mean maximum Standard [Holdaway 1993]) on the basis of surface exposure alone Artefacts dimension deviation may lead to quite variable results depending on the nature of post-depositional geomorphic and pedologic Channel 2 257 38.5 15.9 changes. However, rather than viewing the results pre­ 0-2m buffer 340 36.1 13.7 sented here as a 'cautionary tale', we consider that our 2-4m buffer 292 35.5 14.8 mapping strategy offers the opportunity to control for differential visibility. By excavating small sample areas, 4-6m buffer 351 36.6 13.3 it should be possible to model the degree to which differ­ Artefact mean maximum dimension in Channel 2 and 2, ent land-surfaces are hiding artefacts perhaps providing 4 and 6m buffers from channel. F = 2.25, df= 3, 1236, quantifiable indices to allow true artefact densities to be p = 0.08 estimated. If these densities are to continue to form a part of archaeological significance assessments then further research on the subject needs to be undertaken. Table 3. The effects of water sorting in channels.

The effect of water movement on artefact horizontal integrity Lag Surface unit immediately adjacent and second, in an independent study, the hypothesis of water movement As discussed above, the Lag surfaces at TIB 13 are con­ affecting artefact distributions across the site as a whole sidered to be the result of sheet erosion caused by the was tested and rejected (Pigdon 1997:102). action of rainsplash and water run-off on land subject to The rationale for comparing artefacts in Channels to overgrazing. While this erosion is essential to the creation those on bordering Lag Surface was based on the assump­ of an archaeological record amenable to the recovery sys­ tion that Channels mapped at TIB 13 were a recent feature tems discussed here, there remains the possibility that this formed by concentrated runoff. The artefacts they contain erosion has substantially affected the location as a whole, should, to some extent, show the effect of water transport, moving artefacts horizontally to such an extent that the particularly size sorting (Schick 1987), since small arte­ patterns in their distribution relate more to recent erosion facts will tend to be differentially removed in regions sub­ than to processes connected with use. Clearly, resolving ject to rapid water flow. If the effects of water transport the degree to which water has affected the location is crit­ were limited to Channels we would expect the effects of ical because the viability of the project as a whole rests size sorting to be absent from neighbouring Lag Surface. on the existence of an archaeological record that has been Only if the whole site were size sorted would this test fail lagged i.e. is recognised to be vertically displaced, but to indicate the effect of water movement, a possibility retains much of its horizontal integrity. assessed in the second set of tests below. The question of movement was addressed in three The test for the effect of artefact movement within ways. First, a lower size limit was imposed (20mm in Channels compared to those found distributed along their maximum dimension) below which artefacts were not banks used the Buffer command in the GIS software to recorded. This was based on experimental work by subset groups of artefacts at given distances from the Schick (1987:96) that suggested artefacts smaller that edges of Channels. Individual rills and channels within this dimension were particularly susceptible to move­ the Channel coverage varied considerably in their dimen­ ment through sheetwash erosion. Artefacts larger than sions (Figure 2); many having widths less than 30cm. 20mm were moved short distances by water but then Because these rills and channels were small, they con­ tended to become stationary with time. Two tests were tained very few artefacts making it difficult to detect size then undertaken: first, artefacts present within the sorting by water flow. A few of the rills and channels Channels landform unit were compared with those on the were larger, and in these there were sufficient artefacts to

34 HoldawayetaL(1998) Chapter 1: Introduction allow meaningful tests. Table 3 gives the results of .2500 buffering experiments undertaken on two of the largest channels (Figure 2; Channel 1 area = 903m2; Channel 2 area = 989m2). For each, artefacts falling within the channels and those located within 0-2m, 2-4m and 4-6m buffers constructed on either side of the channels were compared in terms of the mean maximum dimen­ sion for all stone artefacts. For Channel 1 there is a sig­ nificant difference in the mean dimension of artefacts from each of the buffers, with artefacts falling inside the channel having a slightly greater maximum dimension than those artefacts bordering the channel. For Channel 2 the same difference in the means is apparent, but the results of the ANOVA test have a two tailed probability i—i—i—r—T 1 1—r slightly greater (p=0.08) man the accepted cut-off point 7 8 9 10 11 12 13 14 15 These results are consistent with water flow moving rela­ Intersecting groups of buffers ranked by area tively small artefacts, and therefore increasing the over­ 2500 all mean maximum dimension of artefacts within the channel. But there is little evidence for size sorting at increasing distances away from these channels. Since the channels are almost certainly recent (last ISO years) fea­ tures in this landscape we may assume that artefact depo­ sition in prehistory had a negligible effect on the distrib­ ution of artefacts in the channels and on their banks. There can be no doubt that the artefacts on the lagged surfaces have moved to some degree (through deflation), but our findings suggest that only in the larger channels is this movement sufficient to be detectable as size sort­ ing. It seems likely that as Schick (1987) observed, once the 20mm size cut-off is exceeded, artefacts do not con­ tinue to move unless subjected to flow conditions nor­ Intersecting groups of buffers rankedb y area mally only found in channels. Similar results were obtained in an independent study 2500 designed to test for water activity across the location as a whole (Pigdon 1997). A three dimensional model of the TIB 13 surface terrain (a triangular irregular network or TIN) was created using a GIS and the mean length and density of artefacts compared for each contour interval. The results showed no clear patterning across the site as a whole. Only when the lag and channel deposits were tested separately did artefact characteristics show pat­ terning within the channels as predicted by experimental water movement studies (Nash & Petraglia 1987; Schick 1986,1987). In light of the results from these analyses we decided to consider only artefacts from within the Lag Surface 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 landform element since they alone seem to have been Intersecting groups of buffers ranked by area relatively unaffected by post-discard reworking or burial. Figure 3. Ranked buffer areas based on circles of 4m radii drawn around cores and nuclear tools (A), scrapers (B) Assemblage definition and notches (C). The ranks below 15 have been omitted.

As discussed above, we wished to develop a means of analysing assemblages across space independent of the of our study area. Lag surfaces with high visibility and site concept, and because of our use of artefacts as the horizontal integrity are interspersed with channels and minimal depositional and analytical unit, we were drawn sediment islands. As noted above, controlling for differ­ to the intrasite spatial analysis literature (albeit from a ential visibility and integrity required that certain surfaces regional rather than site based perspective). We were, be omitted from analysis. We could not use techniques however, mindful of the geomorphological characteristics that demanded a continuous distribution, nor could we

35 Chapter 1: Introduction Holdaway et al. (1998)

100 m J

Figure 4. Core assemblages based on circles of 4m radius.

use hierarchical and non-hierarchical cluster techniques need to define sites, and a method that allowed us to that grouped artefacts together on the basis of the exis­ incorporate the discontinuous nature of the artefact expo­ tence of breaks in the artefact distribution. To have done sure. These are aspects of LDA that set the technique so would have given priority to landsurface features that apart from approaches proposed by other researchers (eg. are the product of post-European geomorphological Ebert 1992; Wandsnider 1996). Our use of LDA is simi­ processes. This problem also restricted the utility of the lar to Johnson's in that it employs circular sampling units raster based techniques adapted from image analysis that centred on a given artefact category, and varying radius have recently been proposed (Wandsnider 1996). lengths to explore different scales of spatial patterning. Fortunately, in reviewing the literature on intrasite When circular sampling units are laid out on a map of spatial analysis (see, for instance, Blankholm 1991) we the study area, density patterns of the selected category discovered a technique suitable for our purposes, a vari­ can be readily observed. Overlapping sampling units rep­ ant on Local Density Analysis (LDA — Johnson 1984). resent areas where more than one artefact is found within Johnson's original formulation of this technique r metres. Where dense clusters exist, several units will involved establishing the local density of a category of overlap. objects by counting the number of those objects within a Cluster patterns of a single artefact type may be of circular sampling unit, radius r, centred on an object interest in their own right,bu t they can also help to pro­ from a second category. The resulting measure had two vide a framework for investigating more complex pat­ uses: it was used to calculate an index of association terns of assemblage composition. To this end several key between the two categories, and it also could be used as artefact categories, such as cores and common tool types, an aggregation indicator where the first and second cate­ were chosen. Each formed the basis for a separate analy­ gories were the same. sis of the study area. A radius length was selected, and For our purposes LDA offered a flexible method for circular sampling units were mapped. Overlapping units defining assemblages in open site scatters without the were combined to form a single unit, so that the perime-

10

36 Holdawayetal.(I998) Chapter 1: Introduction

Figure 5. Notch assemblages based on circles of 4m radius.

ter length and area of the units varied. The values for used for assemblage definition. The goal then becomes perimeter and area were then plotted on simple line to make sense of the distribution of these assemblages graphs and the point of inflection used to determine the across space based on what we know of stone artefact most significant clusters. production, use and discard in the past Following our interest in examining assemblage com­ position in terms of artefact manufacture, use and dis­ card, the significant cluster zones were used to define TIB 13 Assemblage composition assemblages and these were analysed statistically using a range of technological and typological variables. Each As an illustration of the method, four metre buffers were assemblage had a common characteristic, in mat its arte­ created around three of the most frequent artefact types facts were spatially associated with high-density concen­ from TIB 13: Cores, Scrapers and Notched Tools (see trations of a given artefact category. The object of the Table 1 for definitions). The areas of the resulting clusters analysis was to identify the other ways that the assem­ of buffers were calculated using the GIS software and are blages were alike, and the ways that they were signifi­ graphed in Figure 3. In each case there is a clear point of cantly different This procedure was repeated several inflection in the graphs suggesting the definition of three times, using different radius lengths and men different larger assemblages for Cores and Notched Tools together artefact types as the circle centroids. with a large number of smaller assemblages, and either This use of LDA allowed us to search for pattern two or four larger assemblages for Scrapers. These assem­ across space skipping regions where the artefact distribu­ blages, together with an assemblage formed from all other tion was disturbed as long as these regions fell within the circles combined are analysed here (the four assemblages radius dimension selected. As demonstrated below, in have been selected for scrapers — see Figures 4,5, and 6). searching for interpretable spatial patterns assemblage The results are then compared to determine what generali­ composition is largely invariant no matter what the scale sations at the 4m scale can be determined for TIB 13.

11

37 Chapter J: Introduction HoldawayetaL(1998)

Figure 6. Scraper assemblages based on circles of 4m radius.

Assemblage distribution Assemblage 3) north of the stream channel. Tests based on the frequency of artefact forms confirm that assem­ Although assemblages were created by centring circles blages that share a common spatial location are statisti­ on different tool forms, there are similarities in the cally indistinguishable while significant differences spatial distribution of a number of the assemblages. occur among assemblages that occur in different parts of Thus, in the north-central region of the site Scraper TIB13(Table4). Assemblage 1 and Core Assemblage 1 are spatially quite close. Three assemblages. Scraper Assemblage 2, Notch Assemblage 2 and Core Assemblage 2 share a common Assemblage comparison South Central Location, while Scraper Assemblage 3 and Notch Assemblage 3 are located in the southern part of These general patterns may be refined by analysing the the site. In contrast, Notch Assemblage 1, Scraper nature of the artefacts that intersect with the assemblages Assemblage 4 and Core Assemblage 3 are isolated in the in terms of simple technological and typological attrib­ sense that they do not intersect substantially with any of utes (Tables 5 & 6). In Figure 7 several technological the other major assemblages (although they do intersect indices are graphed that show a largely consistent north- with some of the smaller groups of buffers). This distrib­ south trend in assemblage composition across TIB 13. ution corresponds to patterns in the distribution of the The southern assemblages show relatively high values artefacts selected for buffering. The Notched based for the ratio of tools to complete flakes, but relatively assemblages are distributed across the southern part of low values for the proportion of blades in relation to all the site, as are the scraper based assemblages except that other flake forms, and a low proportion of complete to uniquely Scraper Assemblage 4 is located to the west, broken flakes. The ratio of cores to complete flakes and the Core based assemblages are distributed across shows relatively high values in the south-central and the whole site including an assemblage (Core southern regions with the highest values in the north-

12

38 Holdaway et al. (1998) Chapter!: Introduction

Assemblages Chi-square d.f. phi defined with: Cramer's V

Complete tool vs. Complete flake Core 56.77 3 < 0.001 0.13 Scraper 42.27 4 < 0.001 0.11 Notch 35.13 3 < 0.001 0.12

Complete, broken, proximal, distal flakes Core 47.67 9 < 0.001 0.10/0.06 Scraper 32.20 12 = 0.001 0.09/0.05 Notch 34.37 9 < 0.001 0.10/0.06

Flake form, blade, expanding, intermediate Core 80.41 6 < 0.001 0.14/0.10 Scraper 117.03 8 < 0.001 0.17/0.12 Notch 77.77 4 < 0.001 0.16/0.11

Complete flake to core ratio Core 35.26 3 < 0.001 0.1 Scraper 15.06 4 = 0.005 0.07 Notch 11.53 3 = 0.01 0.07

Core types, bifacial, microblade, unifacial Core 26.45 6 <0.001 0.29/0.20 Scraper1 24.75 6 <0.001 0.31/0.22 Notch 18.89 6 = 0.004 0.31/0.22 Scrapers and notches Core2 5.28 2 = 0.07 Scraper 15.96 4 = 0.003 0.20 Notch 4.76 3 = 0.19

Scrapers, notches, points, adzes and utilised Core 24.96 8 = 0.002 0.21/0.15 ScrapeH 36.96 8 < 0.001 0.27/0.19 Notch 29.21 9 = 0.001 0.25/0.14 1. Scraper 4 not included due to low number of cores (n=4). 2. Core 3 not included due to low number of scrapers and notches (n=5). 3. Scraper 3 and Scraper 4 not included due to zero frequency for points and adzes.

Table 4. Chi-square values for technological comparisons between assemblages defined around cores, scrapers and notches.

west and low values in north-central, central and central- portion of notches occur in the assemblages (Notch 2 west regions. The southern and south-central assem­ and Notch 3) that flank Notch 1. The proportion of blages show relatively low proportions of microblade scrapers reverses this pattern, with the highest propor­ cores compared to cores of other forms, while north-cen­ tions in assemblages toward the centre of TIB 13. The tral and central assemblages show high proportions of highest proportion of scrapers occurs in Scraper 4 in the this core type. central-west. Points and adzes show low frequencies, Tool forms show a more complex pattern (see Table but proportionally they occur more frequently in the 6). There is a clear north-south distribution in the pro­ north than the south (except for Notch 3 in the south portion of notched tools, with southern assemblages where the proportion of adzes is 12.8%). Tools classi­ showing the highest proportion of notched tools fied as Utilised are proportionally more frequent in the (Scraper 3,45.0%; Notch 3,48.7%) and the central-west central regions. assemblage Scraper 4 recording only 12.5%. Among the Taken together, these results show patterns in the asso­ southern and central notch assemblages the highest pro­ ciation of artefacts that largely make sense functionally

13

39 Chapter I: Introduction Holdaway et al. (1998)

Tool and flake fracture Flake morphology Core morphology class frequency frequency frequency T 1-3 U ft, Core Unifai Core Comp Micro Tool Comp Tool Flake Brokei Flake Distal Flake Brokei ft. ». Blade Expan •a. s Assemblage North-West Core 3 4 5 114 53 17 15 20 35 60 12 4 9

North-Central Corel 119 42 1227 228 161 220 212 318 1111 36 28 36 Scraper 1 109 42 983 162 126 172 179 213 924 22 23 25

Central Notch 1 53 21 324 63 57 90 42 122 341 5 8 12

Central-West Scraper 4 13 4 84 19 10 14 17 27 78 1 1 3

South-Central Scraper 2 182 77 787 143 147 143 70 367 780 31 7 59 Notch 2 162 70 700 132 138 131 60 327 696 25 9 51 Core 2 159 83 717 140 142 141 64 341 718 29 9 55

Southern Scraper 3 30 10 121 27 32 21 11 40 145 9 1 7 Notch 3 28 11 130 28 31 23 12 41 146 9 1 6

Table 5. Artefact frequencyfo r technological attributes by assemblage.

Area Assemblage Scraper Notch Point Adze Utilised

Nth-W Core 3 3 (33.3) 2(2Z2) 1 (11.0) 0 3 (33.3) Nth-Cen Corel 62(39.0) 24(15.1) 10(6.3) 27 (17.0) 36 (22.6) Nth Ceo Scraper I 67(45.0) 19(12.8) 9(6.0) 26 (17.4) 28 (18.8) Cen Notch 1 18(24.7) 22(30.1) 3(4.1) 8(11-0) 22(30.1) Cen-W Scraper 4 12(75.0) 2(12.5) 1 (6.3) 0 1 (6.3) Sth-Cen Scraper 2 93 (36.2) 50(19.5) 6(2.3) 24(9.3) 84 (32.7) Sth-Cen Notch 2 70(30.2) 56(24.1) 6(2.6) 20 (8.6) 80(34.5) Sth-Cen Core 2 69(30.8) 49(21.9) 6(2.7) 21 (9.4) 79(35.3) Sth Scraper 3 13(32.5) 18(45.0) 0 3(7.5) 6(15) Sth Notch 3 10(25.6) 19(48.7) 0 5 (12.8) 5 (12.8)

Table 6. Tool types: number (percentage) for each assemblage

14

40 Holdawayetal.(J998) Chapter 1: Introduction

tool • flake A and technologically. Southern and central assemblages 100 have higher proportions of tools with more notches and utilised edges. Technologically they contain a higher pro­ portion of cores, but not of the microblade variety. To the north, there are lower proportions of tools overall, but higher proportions of scrapers, backed blades and adzes compared to other types. Blades are relatively more fre­ quent as are microblade cores. These patterns become clearer if new assemblages rep­ resenting the intersection of the buffered polygons are cre­ ated (Figure 8). Three new assemblages, North Central, South Central and Southern are created through intersec­ tion, with 1511,1397 and 252 artefacts respectively. Table 7 summarises chi-square values for a series of technologi­ * 10 cal comparisons. There are significant, and frequently strong, differences between the North Central assemblages and those in the South Central and Southern parts of the site. There are proportionally more tools and cores in the southern parts of the site and more blades together with more microblade cores in the north. Comparisons of tool 70 type frequency show more scrapers, points and adzes to the north and more notches to the south.

* 60 Discussion

TTB13 is only one location, and it would be dangerous to generalise too much given such a limited spatial sample. llllllllll On the other hand, based on the data available, our 25 analysis using a modified form of LDA has allowed us to detect a significant spatial partem in terms of assemblage variability. In the southern portion of TIB 13 we have assemblages that suggest general core reduction together with artefact manufacture involving tools that lack heavy retouch. To the north, adzes and scrapers were discarded more frequently, possibly to work wood, and we have evidence for the knapping and production of backed llllllllll blades possibly to manufacture projectiles. There are, of course, a number of explanations that could be offered for the spatial variation we have detected. We might, for instance, seek a direct ethno­ graphic analog based on accounts of Aboriginal camp structure at European contact. But before we draw such inferences we need to consider that we are dealing with a statistically derived pattern in the distribution of artefacts at a location that was almost certainly occupied more than once. As discussed above, the method for analysing open sites proposed here is not built around direct ethno­ graphic analogies for patterns we are able to detect — even if these patterns seem to fit perfectly with such explanations. The pattern that we have identified at TIB 13 will only become interpretable once additional areas have been recorded and this pattern, or others, Figure 7. Technological comparisons by assemblage repeated. Eventually we may be able to comment on ranked north to south according to their relative location. stone procurement, use and discard and suggest why A. Flake and tool proportions, B. blades as a percentage some locations retain a patterned distribution of artefacts of total flakes, C. complete flakes as a percentage of and some do not But for the moment demonstrating the all flakes, D. core to complete flake proportion, existence of significant patterns in the composition of E. microblade cores as a percentage of all cores. spatially distinct assemblages is sufficient.

15

41 Chapter 1: Introduction Holdaway et al. (1998)

Chi-square d.f. p phi Cramer's V

Complete tool vs. Complete flake 23.01 2 < 0.001 0.11 Complete, broken, proximal, distal flakes 11.63 6 =0.07 Flake form, blade, expanding, intermediate 80.48 4 < 0.001 0.19/0.13 Complete flake to Core ratio 1 S.42 2 < 0.001 0.1 Core types, bifacial, microblade, unifacial 20.78 4 < 0.001 0.37/056 Scrapers and notches 1S.3S 2 < 0.001 0.28 Scrapers, notches, points, adzes, utilised 41.30 6 < 0.001 0.35/0.25

Table 7 Technological frequency comparisons between the intersection assemblages.

Chronology archaeomagnetic determinations for die numerous hearth or oven features tiiat are associated with die stone arte­ We have only a limited idea at present when the artefacts facts in die locations where we are working. Excavation at TIB 13 were deposited. Given die typology of the tools of several hearths at a second study location (some dis­ (backed blades and tula adzes), it is likely that they are tance from TIB 13) within Sturt National Park, under­ no older than the mid-Holocene. With the application of taken with permission from the NSW NPWS and the modern dating techniques we will be able to refine this Wangkumara people, reveals in many cases lenses of somewhat, but it is unlikely that we will ever be able to charcoal beneath heat cracked stones. A pilot series of determine the number of times the location was used in ten radiocarbon determinations indicates that we can prehistory. Instead we plan to treat chronology in a simi­ obtain sufficient charcoal for conventional dates. lar way to our analysis of artefacts — to search for inter- Archaeomagnetic dating of these same hearths will pro­ pretable pattern that is the result of multiple behavioural vide a check on die radiocarbon determinations and aid events. At the time of writing, we are only able to report in determining the degree of displacement of hearth the techniques we intend to apply in our research at material since they were in use. During the course of die Sturt, although research along all the lines discussed here project we plan to obtain a large number of additional has begun. samples from hearths both exposed and buried at our We plan to deal with chronology in two ways. First, we study location. Exposed hearths are pedestaled on eroded intend to develop a landscape chronology by identifying surfaces but die heat cracked stones have acted to main­ sedimentary units within a catchment and dating those tain die integrity of die hearth and to protect the charcoal units. This will allow us to determine the maximum and from erosion. We plan to use a magnetic susceptibility minimum ages for the sedimentary sequence currently meter to detect buried hearths. When fitted with a ground preserved in the valley fill. We will use this chronology to search loop, the susceptibility meter can be used to suggest a maximum age for the artefacts currently detect concentrations of ferrimagnetic minerals that exposed on the surface. In effect, the sediment chronol­ occur in hearths a few centimetres below the ground ogy will give us a resolution similar to the occupation (Thompson & Oldfield 1986). periods frequently defined by archaeologists working on We need to date a large number of hearths because die stratified deposits; an indication of die general time frame archaeological chronology relies on detecting patterns in which artefacts were deposited without the identifica­ among large numbers of determinations. Because the tion of specific oppositional events. Studies undertaken so stratigraphy has been disturbed by erosion, we cannot far indicate die feasibility of constructing a sedimentary associate artefacts with particular dates, but we can look chronology by obtaining charcoal for radiocarbon deter­ at the range of values for determinations across space. minations and quartz sands for optically stimulated lumi­ Use of a particular location at one time will result in nescence (OSL) from the sediments in the valley. Of more hearths dating to the same period; use of a location these, OSL has die potential to produce a more detailed through timewil l produce hearths witii a spread of dates. chronology since it is better suited to alluvial sediments We are less interested in die magnitude of single points than thermoluminescence, has a broad temporal range (70 in time, rather we seek clusters of dates tiiatma y corre­ years to >800 k years), and is not limited by die availabil­ late with patterns that we can detect in die distribution of ity of charcoal (Murray et al. 1995). As only a small sam­ stone artefacts. One cluster of dates with a particular ple is required, die chances of modern contamination via assemblage type will not be very useful since it is possi­ post-depositional processes such as bioturbation can be ble that die artefacts and die hearths were deposited at minimised by careful selection of die samples. different times. A repeated association, on the other Our second method involves the definition of an hand, will reduce the probability that dates and artefacts archaeological chronology by obtaining radiocarbon and are present in die same location by chance.

16

42 Holdawayetal.(1998) Chapterl: Introduction

North central intersection assemblage South central intersection assemblage Southern intersection assemblage

so 100 m i 1

Figure 8. Regions where assemblages defined with 4m circles intersect

Raw materials talline quartz matrix with a coarse texture to an amor­ phous, fine textured material (Sullivan & Simmons At TIB 13 raw material was identified as silcrete, quartz, 1978:56; Watts 1978). A total of 35 silcrete sources have and quartzite (Table 1). been located and described in terms of material type, Since this location was recorded, Ph.D. research (TD) colour, lithology, outcrop form, production, size and den­ has commenced focusing on identifying the sources of sity (Watts 1978; Hiscock & Mitchell 1993). The majority lithic raw material represented in the archaeological of the sources are silicified from a sandstone with a record of the region. An area stretching from the medium to fine texture (25 sources), and outcrop as Thomson's Creek to the Twelve Mile Creek Gorge and weathered boulders, while localised inclusions of silicified across to the township of Tibooburra has been surveyed siltstones form fine, amorphous silcrete (eight sources). (Figure 1) revealing four categories of abundant, high The quality, size and type of outcrop has affected the den­ quality lithic material: silcrete, quartz, quartzite and sity and extent of reduction with larger, high quality homfels. Silcrete occurs along the Tertiary capping of sources tending to have more debitage. These larger the Mt. Wood Ranges or within the stream catchments sources conform with Wilke & Schroths' (1989) definition exposed through erosion. In contrast, homfels occurs of quarries and will be used to research on the nature of only as an isolated metamorphic/igneous island near the direct procurement strategies undertaken at the sources. Tibooburra township (Figure 1). Within this zone, nod­ ules of quartzite and milky quartz are found on gibber plains. The Mt Wood Ranges also have limited amounts Expected outcomes of quartz nodules found within a mainly silcrete gibber. The silcrete can be divided into two main and four sub Both the research reported here, and that outlined for the groups based on hand specimen variation in texture, frac­ future, will allow us to construct a model of the way arte­ ture and matrix of the material, and ranges from a crys­ fact assemblages were deposited at different locations and

17

43 Chapter 1: Introduction HoldawayetaL(1998)

times within the landscape taking into account the nature pretable. We view this result as encouraging and it gives of archaeological surface exposure. Our geomorphologi- us the confidence to extend the research program to cal research will allow us to characterise these locations investigate additional areas in the hope of identifying in terms of their sedimentary history and allow some repeated pattern in time and space. The technology we degree of palaeoenvironmental reconstruction. For each have employed enables us to work at a much larger location we will be able to offer an interpretation of the scale, both spatially and in terms of the number of arte­ way artefacts were abandoned related to the possible sig­ facts recorded, compared to many other archaeological nificance of artefact associations related to stone procure­ projects in Australia. Theoretical works on hunter-gath­ ment and artefact manufacture, use and discard. In terms erer site formation tell us that we must think in spatially of writing prehistory, the points of most interest will extensive areas. In Australia we are fortunate in having begin to appear as more locations are investigated and a very large exposures available covered with tens of thou­ variety of models for artefact abandonment developed. sands of artefacts. This record has great potential — we The sophistication of our explanations will develop as we hope that our research will stimulate other groups to identify locations where our models should fit — but begin working on this record. apparently don't It will then be up to us as prehistorians to provide explanations for these anomalies. Clearly this is a long-term research endeavour. We are Acknowledgments not searching for a single predictive model that will allow the archaeological record to be interpreted in all Field work at TIB 13 was undertaken during June 1995 places. But we do consider that the research we are and April 1996 with accommodation and logistical sup­ undertaking provides one way around the impasse cur­ port provided by the NPWS Tibooburra District Office. rently faced by many archaeologists dealing with surface There was close liaison with the Tibooburra Local artefact scatters. It is well recognised by those working Aboriginal Land Council and the Wangkumara Tribal with such material that the significance of this type of Council to ensure that they were satisfied with our non­ record is best assessed in terms of a regional settlement invasive methods. Financial support was provided by a model (Holdaway 1993). What has been lacking until Central Starter grant, La Trobe University to SH, a now is not so much the models, but a set of methods that Macquarie University Research Grant to PF, an will allow these models to be truly tested against spa­ Australian Institute of Aboriginal and Torres Strait tially defined artefact assemblages. The methods dis­ Islander Studies small grant to SH and a Collaborative cussed in this paper suggest one way that we may begin Australian Research Council grant to SH and DW. to search for the types of patterns that should be there Funding for equipment was provided by the Standing according to the theoretical literature. Committee on the Allocation of Equipment Funds, La The other major outcome is the application to the con­ Trobe University through a grant to SH. Versions of this servation of the archaeological record. It is now possible paper were presented at seminars in the Department of to see what the distribution of archaeological materials Paleoanthropology and Archaeology, University of New over a landscape is like. In management archaeology it is England, and in the Departments of Anthropology at the very difficult to conceptualise what it is that we are trying University of Pennsylvania, University of Michigan, to protect since the conditions of exposure and ground University of Utah and the Bishop Museum. We thank surface visibility are normally very patchy and the Caroline Bird, Brad Dilli, Tim Murray, Peter Mitchell, observable artefact assemblages difficult to interpret Jim Rhoads, Robin Torrence, LuAnn Wandsnider and Sub-surface testing is hampered by the lack of knowledge Peter White for critical comments on drafts of this paper. of what sample sizes are appropriate (Wobst 1983) so the Badger Bates, Phillip Kerwin, Cecil Ebsworth, Ainsley conservation of open site archaeology is frequently little Ebsworth, Daphne Monaghan, Jenny Tulloch, Denis more than guesswork. The methodology described here Gnaden, Rita Olsudong, Bill Kerse, Rudy Frank, Ming shows that the patterned distributions of cultural evidence Wei, and Philip Purcell helped with the fieldwork.Min g over the landscape are much more extensive than usually Wei drew the figures. considered in management practice.

Conclusion References

The location discussed in this paper is not particularly Blankholm, H. 1991. Intrasite Spatial Analysis in Theory and large on a landscape scale, but at just under 30,000m2 it Practice. Aarhus University Press, Aarhus. Cameron, C. & S. Tomka 1993. Abandonment of Settlements is probably as large as any in Australia with near contin­ and Regions: Ethnoarchaeological and Archaeological uous visibility to have been analysed on an artefact by Approaches. Cambridge University Press, Cambridge. artefact basis. With 10,000 artefacts we have a reason­ Carr, C. 1984. The nature of organization of intrasite archaeo­ ably large database, and by analysing this spatially, we logical records and spatial analytical approaches to their are able to detect a pattern in the distribution of artefacts investigation. Advances in Archaeological Method and at an assemblage level that may be functionally inter- Theory 7:103-222.

18

44 Holdaway et al. (1998) Chapter 1: Introduction

Cosgrove, R. 1995. The Illusion of Riches: Scale, Resolution In E. Kroll & T. Price The Interpretation of Archaeological and Explanation in Tasmanian Pleistocene Human Spatial Patterning. Plenum Press, New York. Pp 61-76. Behaviour. Tempus Reparatum: British Archaeological Pain C.F., Chan, R., Craig, M, Hazell, M., Kamprad, J. and Reports, International Series 608, Oxford. Wilford, J. 1991. RTMAP: BMR Regolith Database Field Dunnell, R. 1992. The Notion Site. In J. Rossignol & L. Handbook. Bureau of Mineral Resources Record 1991/29, Wandsnider (eds.) Space, Time and Archaeological Canberra Landscapes. Plenum Press, New York. Pp 21-41. Pigdon, D.J. 1997. GIS and terrain modelling in open site Dunnell, R. & W. Dancey 1983. The siteless survey: a regional archaeology. Identifying fluvial processes in an open site scale data collection strategy. Advances in Archaeological from north-western New South Wales. Unpublished Master Method and Theory 6:267-287. of Applied Science thesis, University of Melbourne. Ebert, J. 1992. Distributional Archaeology. University of New Rossignol, J. & L. Wandsnider 1992. Space, Time and Mexico Press, Albuquerque. Archaeological Landscapes. Plenum Press, New York. Fanning, P.C. 1996. Stream-side erosion: implications for Schick, K.D. 1986. Stone Age Sites In The Making: future rangelands management, Proceedings of the 9tn Experiments In The Formation And Transformation Of Biennial Australian Rangelands Conference, Port Augusta, Archaeological Occurrences. BAR International series 319 September 1996. Pp.231-232. Schick, K. 1987. Experimentally derived criteria for assessing Freeman, L. & K. Butzer 1966. The Acheulian station of hydrologic disturbance of archaeological sites. In D. Nash Torralba (Spain). A progress report Quaternaria 8:9-22. & M. Petraglia (eds.) Natural Formation Processes and the Hiscock, P. & S. Mitchell 1993. Stone Artefact Quarries and Archaeological Record. BAR International Series, Oxford, Reduction Sites in Australia. Government Publishing 352. Pp86-107. Service, Canberra. Schiffer, M. 1972. Archaeological context and systemic con­ Holdaway, S. 1993. Archaeological Assessment Standards and text. American Antiquity 37:156-165. Methodological Design for the Hunter Valley. Report to the Sullivan, M. & Simmons, S., 1978. Silcrete: A classification New South Wales National Parks and Wildlife Service, Sydney. for flaked stone assemblages. The Artefact 4:51-60. Johnson, 1.1984. Cell frequency recording and analysis of arti­ Thompson, R., & Oldfield, F., 1986. Environmental fact distributions. In H. Heitala (ed.) Intrasite Spatial Magnetism. Allen & Unwin, London. Analysis in Archaeology. Cambridge University Press, Wandsnider, L. 1996. Describing and comparing archaeologi­ Cambridge. Pp 75-96. cal spatial structures. Journal of Archaeological Method McPherron, S. & S. Holdaway 1996. Entrer Trois. In H. Dibble andTheory 3:319-384. and S. McPherron A Multimedia Companion to the Middle Watts, S. 1978 Nature and occurrence of silcrete in a portion of Paleolithic Site of Combe-CapeUe Bas (France). CD-ROM, inland Australia. Unpublished Ph.D. thesis, University of University of Pennsylvania Museum, Philadelphia. Sydney. Murray, A. S., Olley, J. M. Caitcheon, G. G. 1995. Whallon, R. 1973. The spatial analysis of Palaeolithic occupation Measurement of equivalent doses in quartz from contempo­ areas. In C Renfrew (ed) The Explanation of Cultural Change: rary water Iain sediments using optically stimulated lumi­ Models in Prehistory. Duckworth, London. Pp 115-30. nescence Quaternary Science Reviews 14:365-371. Whallon, R. 1974. Spatial analysis of occupation floors II: Nash, D.T. & M.D.Petraglia 1987. The impact of fluvial Application of nearest neighbour analysis. American processes on experimental sites. In D.T.Nash and Antiquity 39:16-34. M.D.Petraglia (Eds) Natural Formation Processes and the Wilke, P. & Schroth, A. 1989. Lithic raw material Prospects in Archaeological Record. BAR International series 352, the Mojave Desert, California. Journal of Califomian and Pp.108-130 Great Basin Anthropology 11:147-174. NSW NPWS 1997. Minimum requirements and standards for Witter, D. 1992. Regions and Resources. Unpublished Ph.D. archaeological survey methodology and significance assess­ thesis, Australian National University. ment in cultural resource impact statements. Unpublished Wobst, M. 1983. We can't see the forest for the trees: sampling report. National Parks and Wildlife Service, Hurstville. and the shapes of archaeological distributions. In J. Moore O'Connell, J. 1987. Alywara site structure and its archaeologi­ and A. Keene (ed.) Archaeological Hammers and Theories. cal implications. American Antiquity 52:74-108. Academic Press, New York. Pp. 37-85 O'Connell, J., K. Hawkes, & N. Blurton Jones 1991. Yellen, J. 1977. Archaeological Approaches to the Present: Distribution of refuse-producing activities at Hadza residen­ Models for Reconstructing the Past. Academic Press, New tial base camps: implications of archaeological site structure. York.

19

45 Chapter 1: Introduction Holdaway et al. (1998)

46 Chapter 1: Introduction

Johnson, I. & North, M. (1997). Archaeological Applications of GIS. Proceedings of Colloquium II, UISPP XIIIth Congress, Forli, Italy September 1996. Sydney University Archaeological Methods Series 5.

GIS Analysis of Artefact Distributions on an Eroding Landscape: The Western New South Wales Archaeological Project Simon Holdaway, La Trobe University Patricia Fanning, Macquarie University Dan Witter, NSW National Parks and Wildlife Service

1. Aims The aims of the Western New South Wales Archaeological Project are to: • Investigate an archaeological landscape with the artefact as the minimal depositional unit;

• Investigate the surrounding geomorphological landscape in order to control for post- depositional effects, including deflation, water movement and artefact visibility;

• Having controlled for these effects, determine the composition of spatially associated assemblages;

• Investigate the chronology of landscape use, looking for repeated patterns in assemblage Figure 1 composition and landscape re-use; The study area • Develop a predictive model of site re-use, based on landscape attributes, for heritage management purposes.

2. Background Much of the arid and semi-arid parts of western NSW have been significantly affected by erosion following the introduction of sheep grazing in the late 1800s. Topsoils have largely been removed, leaving gravel lags and subsoils exposed at the surface. The eroded material has accumulated in valley floors, filling waterholes with sandy sediments and altering hydrogeomorphic regimes along most of the upland creek systems. The erosion has exposed many thousands of Aboriginal stone artefacts at the surface, allowing investigation of the spatial distribution of the artefacts without the need for excavation. Deflation by wind and water has affected the vertical integrity of the assemblages, but the horizontal integrity appears to have remained largely intact.

3. Macromorphology Identification of landform elements is the basis of geomorphic mapping in the study area. Boundaries between elements are recognised by field reconnaissance and surveyed using GPS and EDM equipment. GIS images are built and displayed using Arc/Info and ArcView. Sixteen Holdaway et al. (1997) Chapter 1: Introduction such landform elements have been identified so far in the Stud Creek catchment in Sturt National Park, on the basis of spatial location, topography and dominant process i.e. whether they are stable (lagged), actively eroding or depositional. Exposure of artefacts is considered to have the greatest potential on lagged and actively eroding elements, and the least on those where deposition dominates (such as channel margins on the valley floors). Areas where runoff is concentrated into rills are also mapped, since the potential for downslope transport of clasts (including artefacts), and hence disturbance of the spatial integrity of assemblages, is highest in these areas. Preliminary results from a pilot study on a site in the area (TIB13) show that, as expected, the mean size of artefacts in rills declines in the downslope direction. A buffered area around all rills will therefore be eliminated from the assemblage analysis.

4. Micromorphology | | Even where erosion is dominant, such as on the Eroding Valley Margin and Valley Floor Lag elements, there are patches of different surface materials which may influence artefact recovery. A micromorphological coverage has therefore been compiled for these elements (Figure 6). Seven different units have been recognised so far; with the exception of 'vegetation', the labels describe the nature of the dominant surface material or regolith within that unit: 1. mud (fine sediment deposition); 2. sand (medium sediment deposition); 3. gravel (coarse sediment deposition); 4. sand & gravel (mixed medium and coarse sediments); 5. cemented gravel (coarse sedimentary material now exposed at the surface by erosion); 6. subsoil (structured clay-rich material now exposed at the surface by erosion); 7. vegetation (usually growing out of sediment mounds; may be grasses, shrubs or trees). Mapping at this level allows a comparison of the way artefacts are clustered or dispersed in relation to post-depositional surface processes.

48 Holdaway et al. (1997) Chapter1: Introduction

Figure 5 Boundaries of the micromorphological (purple) and stone artefact (green) surveys Figure 6 Micromorphological survey

| |

5. Stone artefact analysis All stone artefacts with a maximum dimension of 20mm or above have been located in three dimensional space with EDM equipment, c. 24,000 pieces within an area of just 0.045 sq. km (Figure 7). Attributes are recorded for each artefact by logging into palmtop computers, and include a range of typological and technological attributes, such as: completeness; flake form; presence and proportion of cortex; platform type; termination; and length, width and thickness dimensions. The material type is also recorded. Data are down-loaded from the palmtops each day and linked to the GIS database in a field computer laboratory. After allowing for post-depositional disturbance, the GIS is used to define assemblages which are compared among each other using univariate statistics. The comparisons allow us to Figure 7 search for patterns related to the technology of Study area for stone artefact analysis, artefact production. They will also allow us to showing 24,000 artefacts as a point coverage identify regions that may have been used for different functions. By relating stone tool assemblages to regolith terrain units, a predictive model of artefact location can be developed. The assessment of the significance of the archaeological record can be made on a regional basis, rather than a site basis, which will facilitate environmental impact assessment of proposed activities, such as mining and infrastructure development.

49 Holdaway et al. (1997) Chapter 1: Introduction

6. Landscape history The landscape evolution in the study area is being investigated at the same time, via stratigraphic and geophysical analysis of valley fills. Together with radiocarbon and archaeomagnetic dating of Aboriginal hearths exposed at the surface, it is expected that these investigations will provide a temporal envelope for landscape usage by Aboriginal people. Work to date indicates that, prior to European occupation, the stream channels contained waterholes lined with muds, which would have held water for considerable periods after rain and hence provided the people with a relatively stable water source. As Figure 8 shows, these muds are now overlain by red sandy sediments derived from erosion of the hillslopes higher up in the catchment.

Figure 8 Section showing sediments eroded from catchment slopes overlying waterhole muds

50 Chapter Two

V J J

/ / m J .y

/ - ar

* ••

Silcrete flakes, cores and tools forming a lag on a surface exposed by erosion.

Surface Stone Artefact Scatters:

Why Can We See Them?

CHAPTER TWO

SURFACE STONE ARTEFACT SCATTERS: WHY CAN WE SEE THEM?

2.1 Introduction

Western NSW is an ideal location to undertake the kinds of geoarchaeological research presented in this thesis. Its current arid climate and related sparse vegetation cover mean that surface artefact scatters are relatively easy to see (Figure 2.1). Moreover, artefact exposure has been enhanced over much of the region by geomorphic dynamics, particularly accelerated erosion that is a consequence of the introduction of sheep grazing in the mid- to late 1800s. A sparsely scattered human population and relatively low intensity pastoral land use mean that much of the archaeological record remains available for examination.

Figure 2.1. Stone artefacts lying on an eroded surface in the Stud Creek study area, Sturt National Park. Depositional surfaces containing sediments which obscure artefacts from view can be seen in the background.

Chapter 2: Why Can We See Them?

52 Chapter 2: Why Can We See Them?

2.2 Recent Geomorphic Landscape Change in Western NSW

In the first of two papers in this chapter (FANNING 1999), I describe geomorphic evidence for landscape change in the last 150 years or so since European occupation in three catchments in western NSW (see Figure 1 in FANNING 1999 for location map). Observations and measurements include topsoil loss and surface lowering using erosion pins at Fowlers Gap Arid Zone Research Station, channel enlargement and knickpoint retreat using repeated measurements at monumented cross-sections along Homestead Creek in Mutawintji National Park, and analysis and absolute dating of valley fill regolith sequences in both these and Stud Creek catchment in Sturt National Park.

An indication of the time of onset of this latest phase of geomorphic landscape evolution in the catchments studied has been in part provided by archaeological evidence. In the paper I describe the typical stratigraphic context in which the remains of Aboriginal heat-retainer ovens are found across the region and the evidence this provides for erosion of topsoil and surface lowering since the ovens were last used. Radiocarbon determinations on charcoal sampled from ovens along Stud Creek (see Chapter Four: HOLDAWAY ET AL. 2002, Table 1) indicate that some were in use as recently as 220±55 y BP. Thus, the surfaces into which the cooking pits were dug were probably relatively intact just prior to the time of European contact, about 150 years ago. I also describe a stratigraphic section from Giles Creek in Mutawintji National Park in which heat-retainer ovens dating to 220±50 and 270±50 y BP have been buried by red sandy sediments. These dates also suggest that the landsurfaces into which the ovens were dug were relatively intact at the time of European contact, and that their subsequent burial by loose red sandy sediments may reflect disturbance of catchment regolith cover when sheep grazing was introduced.

Other researchers describe similar sediment sequences from widespread locations across the region, and indeed across the whole state of NSW (e.g. Crighton 2000, Gore et al. 2000, Jansen 2001, Pickard 1994, Starr 1989, Wasson et al. 1998). These authors present abundant evidence that the red sandy sediments which cap older floodplain deposits (Figure 2.2), variously referred to as ‘post-settlement alluvium’ (PSA) or ‘post-European material’ (PEM), post-dates European occupation. I conclude that the geomorphological and archaeological evidence is overwhelming for the most recent phase of geomorphic landscape change across the western NSW region being post-European i.e. less than 200

53 Chapter 2: Why Can We See Them? years old (FANNING 1999). A model is presented which illustrates the role of geomorphic dynamics, encompassing process/response mechanisms, in determining the type and direction of landscape change in the recent past in western NSW. This then becomes the foundation upon which a landscape-based framework for studying Aboriginal stone artefact scatters is developed in Chapters Three, Four and Five.

2.3 Significance of Landscape Change for Archaeological Investigations of Stone Artefact Scatters

In the second paper (HOLDAWAY ET AL. 2000) the significance of this landscape change for archaeological research into prehistoric Aboriginal occupation of the rangelands of western NSW is outlined. The record of Aboriginal hunter-gatherer activity, i.e. the stone tools they manufactured, used and then discarded needs to be studied in spatially extensive sets. The recent erosion in western NSW described in the previous paper (FANNING 1999) allows archaeologists to undertake such studies because it has, in effect, ‘excavated’ thousands of square metres of surface in any one place, areas far larger than even the most extensive archaeological excavations. Moreover, erosion of the fine sediments surrounding and supporting the stone artefacts has meant that the discard products of Aboriginal activity have been conflated, that is, vertically deflated and concentrated into a single layer. In effect, the artefact scatters have become ‘time-averaged’ (sensu Stern 1994) or ‘trans- episodic’ (sensu Wandsnider 1989) deposits in which the products of many individual behaviours are now concatenated. But rather than confounding attempts by archaeologists to understand human behaviour by analysing lithic scatters, artefact conflation can make spatial discard patterns easier to see, as shown by the results of the pilot study at TIB 13 (HOLDAWAY ET AL. 1998). By looking at the nature of the artefacts accumulated over time in different places in the landscape, archaeologists are able to distinguish those places which were used frequently from those that were used less often, and those used for a larger number and greater variety of activities from those used more sparingly. A ‘place use history’ (sensu Wandsnider 1989) can then be developed.

The paper presents some results of preliminary analyses of artefact data from Stud Creek in Sturt National Park (see Figure 2 in HOLDAWAY ET AL. 2000 for location map) to illustrate these new approaches to surface artefact scatters. The region has a rich lithic resource base, dominated by silcrete occurring both as massive outcrops and pavements of closely packed subspherical cobbles, called ‘gibber’ (Dury 1970). The gibber (Figure 2.3) is

54 Chapter 2: Why Can We See Them?

Figure 2.2. Dan Witter and John Jansen examining the sharp contact between bright red post-European material (PEM) and darker buried floodplain sediments in the sidewall of Homestead Creek at Mutawintji National Park.

Figure 2.3: Gibber pavement in the Stud Creek study area, used as a stone source for tool making by Aboriginal people. Field crew are marking artefact locations with pink flagging.

55 Chapter 2: Why Can We See Them?

56 Chapter 2: Why Can We See Them? derived from concretions within the upper layer of silcrete developed in Early Cretaceous sediments that outcrop along the western edge of the Stud Creek catchment. The lower layer is more massive and columnar, and has been extensively quarried by Aboriginal people (Doelman et al. 2001).

The geomorphic mapping which forms the framework for stratified sampling of the artefact data sets is also briefly described in this paper; a more complete description and analysis can be found in FANNING AND HOLDAWAY (submitted) in Chapter Three.

2.4 References

Crighton, P. (2000). Post-European Sediments in Semi-Arid Western New South Wales: Sinks and Sources. Unpublished B.Sc. (Hons) thesis, Macquarie University, Sydney. Doelman, T., Webb, J. & Domanski, M. (2001). Source to discard: patterns of lithic raw material procurement and use in Sturt National Park, northwestern New South Wales. Archaeology in Oceania 36, 15-33. Dury, G.H. (1970). Morphometry of gibber gravel at Mt Stuart, New South Wales. Journal of the Geological Society of Australia 16, 655-666. FANNING, P.C. (1999). Recent landscape history in arid western NSW, Australia: a model for regional change. Geomorphology 29, 191 – 209. Gore, D., Brierley, G., Pickard, J. & Jansen, J. (2000). Anatomy of a floodout in semi-arid eastern Australia. Zeitschrift fur Geomorphologie Supplemente-Bande 122, 113-139. Jansen, J.D. (2001). Bedrock Channel Morphodynamics and Landscape Evolution in an Arid Zone Gorge: Sandy Creek Gorge, Northern Barrier Range, South-Eastern Central Australia. Unpublished PhD thesis, Macquarie University, Sydney. HOLDAWAY, S., WITTER, D., FANNING, P., MUSGRAVE, R., COCHRANE, G., DOLEMAN, T., GREENWOOD, S., PIGDON, D. & REEVES, J. (1998). New approaches to open site spatial archaeology in Sturt National Park, New South Wales, Australia. Archaeology in Oceania 33, 1-19. HOLDAWAY, S.J., FANNING, P.C. & WITTER, D.C. (2000). Prehistoric Aboriginal occupation of the rangelands: interpreting the surface archaeological record of far western New South Wales Australia. The Rangelands Journal 22, 44 - 57. HOLDAWAY, S.J., FANNING, P.C., WITTER, D.C., JONES, M., NICHOLLS, G. & SHINER, J. (2002). Variability in the chronology of late Holocene Aboriginal

57 Chapter 2: Why Can We See Them?

occupation on the arid margin of southeastern Australia. Journal of Archaeological Science 29, in press. Pickard, J. (1994). Post-European changes in creeks of semi-arid rangelands, “Polpah Station”, New South Wales. In A.C. Millington & K. Pye (eds) Environmental Change in Drylands: Biogeographical and Geomorphological Perspectives, pp. 271-283. Wiley: London. Starr, B. (1989). Anecdotal and relic evidence of the history of gully erosion and sediment movement in the Michalago Creek catchment area, NSW. Australian Journal of Soil and Water Conservation 2(3), 26-31. Stern, N. (1994). The implications of time-averaging for reconstructing the land-use patterns of early tool using hominids. Journal of Human Evolution 27, 89-105. Wandsnider, L.A. (1989). Long-Term Land Use, Formation Processes, and the Structure of the Archaeological Landscape: a Case Study from Southwestern Wyoming. Unpublished PhD thesis, University of New Mexico, Albuquerque. Wasson, R.J., Mazari, R.K., Starr, B. & Clifton, G. (1998). The recent history of erosion and sedimentation on the Southern Tablelands of southeastern Australia: sediment flux dominated by channel incision. Geomorphology 24, 291-308.

58 Publication

Due to copyright laws, the following article has been omitted from this thesis. Please refer to the following link for the abstract details.

Fanning, Patricia C. (1999). Recent landscape history in arid western New South Wales, Australia: a model for regional change. Geomorphology, 29(3-4), 191-209. http://dx.doi.org/10.1016/S0169-555X(99)00014-8

59

60

61

62

63

64

65

66

67

68

69

70

71

72

73

74

75

76

77

78

Publication

Due to copyright laws, the following article has been omitted from this thesis. Please refer to the following link for the abstract details.

Holdaway, Simon J., Fanning, Patricia C. & Witter, Dan C. (2000). Prehistoric aboriginal occupation of the rangelands: Interpreting the surface archaeological record of far western New South Wales, Australia. Rangeland Journal, 22(1), 44-57. http://dx.doi.org/10.1071/RJ0000044

79

80

81

82

83

84

85

86

87

88

89

90

91

92

Chapter Three

Trish Fanning marking the boundary between landsurface units prior to landform and artefact survey.

Geomorphic Controls on Spatial Patterning of the Surface Stone Artefact Record

CHAPTER THREE

GEOMORPHIC CONTROLS ON SPATIAL PATTERNING OF THE SURFACE STONE ARTEFACT RECORD

3.1 Artefact Exposure and Visibility

In the previous chapter, I demonstrated the significant role played by recent geomorphic landscape change in “excavating” valley floor deposits and exposing artefacts abandoned by Aboriginal hunter-gatherers in prehistory. Stone artefacts and the remains of heat-retainer ovens are visible at the surface today because erosion, accelerated by land use change from hunter-gathering to sheep grazing, has removed the finer sediments into which the artefacts were incorporated after abandonment, leaving them as a conflated lag – or “blanket” – lying on landsurfaces across extensive areas of arid western NSW. As a result of valley floor incision, channel widening, knickpoint retreat and channel avulsion associated with these changes, some artefact scatters have been entirely eroded away, while valley floor sedimentation in other places has resulted in artefact burial.

These geomorphic processes operating differentially across the physical landscape are the principal determinants of artefact preservation, exposure and visibility in the study area, and, as a consequence, the principal determinants of the nature of the archaeological record at the time of artefact survey (the archaeological ‘document’ sensu Wandsnider & Camilli, 1992). In this chapter, I discuss the geomorphic controls on artefact exposure and visibility in the Stud Creek study area and how artefact surveys can be designed to accommodate this.

I distinguish between “exposure” and “visibility” as follows. Exposure determines the chance of artefacts being present on the surface, i.e. their ‘discoverability’ (Camilli & Ebert 1992). In the Stud Creek study area, it is a particular function of, amongst other things, contemporary landsurface morphodynamics and geomorphic history, not simply a result of erosion as stated by Hiscock and Hughes (1983). Exposure of artefacts is not uniform across the landscape because, in addition to non-uniform artefact discard by people, morphodynamics and geomorphic histories vary from place to place.

Chapter 3: Spatial Patterning

Surface visibility, on the other hand, determines the chances of artefacts being detected if present (Hiscock & Hughes 1983), and is also not uniform, even across surfaces where there is high artefact exposure. Many consider that surface visibility simply refers to the extent to which the ground surface recorded during the survey was covered in vegetation (e.g. David 1996; Hiscock 1991 in Thorley 1998; Schiffer 1987; Seaman et al. 1988; Thorley 1998) but others such as Wandsnider & Camilli (1992) recognise that it is more complex. For example, fine-grained surface sediments comprising sands and mud may obscure artefacts through burial (e.g. Wandsnider 1989) and bioturbation (e.g. Johnson 1990), but coarse- grained sediments such as gravels and cobbles may enhance their visibility through surface armouring. All of these processes act at a range of spatial and temporal scales.

These themes are explored in detail in the first of two papers in this chapter (FANNING & HOLDAWAY, submitted). New approaches to survey and analysis of surface artefact scatters are described, incorporating stratified systematic sampling using a hierarchy of geomorphic landscape features. While archaeological survey stratification on the basis of landscape is not new, with Butzer in 1982 defining the landscape context of archaeological sites at micro, meso and macro scales, the emphasis here is on morphodynamics rather than the more static and climatically deterministic morphogenetic approach (e.g. Butzer 1972, 1982). Areas of high artefact exposure (Figure 3.1) are targeted using known relationships between geomorphic form, contemporary processes and the recent landscape history previously described (FANNING 1999). Wells (2001) recently developed a similar approach by stratifying the geomorphic landscape according to the relative stability of surfaces.

I also attempted to control for differential visibility by mapping surface condition across the survey area and quantifying the effects of variable surface cover types on artefact density (Figure 3.2). I confirmed the expected relationships between erosional surfaces and high artefact densities, and depositional areas and low artefact densities, but my attempts to precisely quantify these relationships were thwarted by considerable variation at the local level. There is, in fact, dynamic interplay between patterns of artefact visibility as controlled by surface condition and patterns of artefact discard that is very difficult to quantify.

94

Chapter 3: Spatial Patterning

Figure 3.1: Eroding Valley Margin (EVM) geomorphic unit in the Stud 1 study area. Erosion has extended down to the columnar blocky subsoil surface, resulting in maximum artefact exposure. The coloured scale bar is 1 m long.

Figure 3.2: Sandy vegetated hummock on sand/gravel surface in the Stud 2 study area. Microtopographic features and surface condition types like these were surveyed and stored as a separate coverage in the GIS. The scale bar is 1 m long.

95

Chapter 3: Spatial Patterning

96

Chapter 3: Spatial Patterning

Doleman (1992) reached similar conclusions in a study of archaeological materials of the Tularosa Basin in New Mexico, U.S.A. He initially proposed two alternative hypothetical models to explain the relative roles of natural and cultural formation processes in determining the distribution of artefacts on the basin floor. The Holocene Litter Model (HLM) proposes that semi-continuous surface artefact scatters are a product of highly dispersed foraging/extraction activities that result in an archaeological record that is primarily controlled by behavioural processes. In contrast, the Geological Disturbance Model (GDM) proposes that the geomorphic processes of erosion and deposition that formed the modern-day geomorphic features, such as coppice dunes, sandsheets and interdunal deflation areas, have formed aeolian windows, whose placement and size are more or less random. Deflation opens these windows through the aeolian mantle, exposing and concentrating artefacts and other objects too heavy to be carried away by the wind. Because the windows control the exposure of archaeological materials, the GDM proposes that geomorphic processes result in a "highly localised and biased archaeological record" and that the geomorphic processes involved were “…expected to have collapsed and smeared portions of the original archaeological distribution…” (Doleman, 1992: 73). However, analyses of the artefactual data led Doleman (1992: 104) to conclude that “…geomorphic processes…have failed to eradicate behaviourally conditioned spatial patterning in at least some contexts…” In other words, the signature of the original artefact discard behaviour was still recognisable despite the geomorphic influence on differential artefact visibility. As is the case at Stud Creek, no simple relationship between micro-scale geomorphic features and artefact spatial patterning could be found.

In a later study aimed at testing Doleman’s models, Buck et al. (1999) measured the linear correlation between artefact density (n/ha) and a range of microtopographic geomorphic variables, including the density of coppice dunes (n/ha), the surface area of the dunes (m2/ha) and the surface area of the deflation zone or aeolian 'window' (m2/ha). The aim was to determine whether or not geomorphic processes of deflation affected artefact density. Their results showed no linear relationship between artefact density and dune density, and only a weak positive correlation between artefact density and surface area of the dunes. Of greater importance to the testing of the model was the finding that there was no linear correlation between artefact density and the surface area of aeolian windows. A logistic regression analysis was performed to predict the presence or absence of artefacts in a hectare based on the number of dunes in that hectare. The regression coefficients were statistically significant from zero; however, the coefficients were so small that the prediction 97

Chapter 3: Spatial Patterning function is essentially constant. Thus varying the number of dunes produces no marked trend in the probability of finding an artefact in the hectare. The authors concluded that modern geomorphic features appear to be having no marked effect on the distribution of artefacts at a microtopographic scale. However, other studies by Doleman & Stauber (1992) have shown high correlation between surface artefact densities and landform type at the macrotopographic scale in the Tularosa Basin. Thus, scale is a major factor in both the geomorphic influence and expression of the archaeological record in arid environments (Buck et al. 1999; Linse 1993).

3.2 Lateral Integrity of Surface Artefact Scatters in the Stud Creek Study Area

The artefact survey protocols outlined in FANNING & HOLDAWAY (submitted) target landsurfaces with the greatest potential for exposure of the archaeological record. I have also attempted to accommodate differential artefact visibility at the time of the survey by including assessment of surface cover and quantifying its effects on artefact density. Observer variability in recording artefact attributes is also accounted for by applying corrections for errors statistically determined from double analysing a random sample of the recorded artefacts (Gnaden & Holdaway 2000). Finally, the lateral integrity of the visible scatters included in the survey needs to be established before archaeologists can draw behavioural inferences from the spatial distribution of the recorded artefacts.

Archaeologists have long recognised that artefact assemblages are subject to post-discard modification by both cultural (i.e. human-induced) and natural (i.e. non-human) formation processes (e.g. Schiffer 1983). Indeed, some have argued that all archaeological deposits are modified to some extent, except for those instantaneously captured and preserved by rapid burial – the so-called ‘Pompeii Premise’ (Binford 1981). Human-induced disturbance includes scuffing, treadage, deliberate burial, scavenging, reuse and recycling. Natural disturbance encompasses the whole range of geological processes from earthquakes and volcanic eruptions to bioturbation by ground-dwelling invertebrates (see Wood & Johnson 1978, Nash & Petraglia 1987, and Stein 2001 for comprehensive reviews).

Studies of natural formation processes have tended to focus on demonstrating disturbance of artefact deposits (e.g. Lancaster 1986; Petraglia & Nash 1987; Reid & Frostick 1985; Rick 1976; Schick 1986; Shackley 1978), rather than on proving lateral integrity (e.g.

98

Chapter 3: Spatial Patterning

Wandsnider 1989). In Australia, many archaeologists bypassed surface scatters of stone artefacts because they were considered to be too disturbed to be of much use in the quest for archaeological indicators of the earliest habitation site. Actualistic studies of artefact taphonomy (after Hiscock 1985, 1990) from open sites are relatively scarce (but see Cameron et al. 1990; Greenwood 1997; Robins 1999). They are also usually limited in scope, in terms of both the area over which disturbance processes are observed and the length of time they are monitored. For example, Cameron et al. (1990) monitored the movement of artefacts out of ten 1 m squares for three years, and Robins (1993, 1999) monitored the appearance/disappearance of stone artefacts in eight 1 m squares also for three years. Process monitoring of this type is usually limited by the duration of PhD research programs and the tenure of research grants, and the chance of capturing rare, catastrophic events is highly unlikely (but see Cameron et al. 1990). Studies of geomorphic processes in the Australian arid zone have shown that trends towards erosion or deposition on a range of surface types can only be detected by monitoring over much longer periods of time, preferably decades (Fanning 1994).

Rather than trying to monitor contemporary artefact disturbance at the Stud Creek study sites, we were more concerned to find out whether or not the effects of disturbance processes that had occurred prior to our artefact survey were detectable and hence might confound any behavioural analysis of the patterns of artefact distribution over the study area. The second of the two papers in this chapter (FANNING AND HOLDAWAY 2001a) presents the results of these investigations. As a consequence of the pilot study previously conducted at the TIB-13 site (HOLDAWAY ET AL. 1998 in Chapter One), we had already controlled for post-discard disturbance effects in two ways. First, we restricted the artefacts recorded in the surveys to those larger than 20 mm maximum dimension (after Schick 1987), and second, we eliminated from the analysis those artefacts that were located in and immediately adjacent to areas of concentrated surface flow, such as rills, gullies and channels (Figure 3.3). The aim of the investigation was to use known relationships between clast length and hydro-geomorphic processes, especially inter-rill overland flow which is the dominant contemporary geomorphic process over the study area, to test whether any evidence of post-discard movement of artefacts by overland flow could be detected in the spatial distributions of the artefacts we had recorded.

The results indicate that, while artefact size and slope angle (a proxy for hydrologic stream power) are statistically significantly related, the variance in maximum dimension explained 99

Chapter 3: Spatial Patterning by gradient is very low. The significant regression result was probably only detectable because of the very large data set used (n = 17,128). We concluded from this result that the spatial distribution of artefacts in the surface scatters we studied is unlikely to have been significantly altered by post-discard hydro-geomorphic processes like inter-rill overland flow. While the surface artefact scatters in the study area are certainly vertically conflated, and thus unlikely to preserve “living floors” in the short-term, functional sense, their apparent lateral integrity means that meaningful associations of artefacts are likely to be detectable. If a temporal framework for the formation of the archaeological record could be constructed, then the artefact scatters present on the surface today could be used by archaeologists to investigate long-term place use by Aboriginal people.

These issues are addressed in Chapter Four.

100

Chapter 3: Spatial Patterning

Figure 3.3: Shallow watercourse traversing the Stud Creek study area. Areas of concentrated water flow like this are eliminated from the survey by marking and surveying a line around the outer boundary. Any artefacts within the gully system are likely to have been laterally displaced from their original discard location and are therefore not included in the survey.

101

Chapter 3: Spatial Patterning

102

Chapter 3: Spatial Patterning

3.3 References

Binford, L.R. (1981). Behavioural archaeology and the “Pompeii premise”. Journal of Anthropological Research 37, 195-208. Buck, B.J., Steiner, R.L., Burgett, G. & Monger, H.C. (1999). Artifact distribution and its relationship to microtopographic geomorphic features in an eolian environment, Chihuahuan Desert. Geoarchaeology: an International Journal 14(8), 735-754. Butzer, K.W. (1972). Environment and Archeology: an Ecological Approach to Prehistory, 2nd edition. Methuen: London. Butzer, K.W. (1982). Archaeology as Human Ecology: Method and Theory for a Contextual Approach. Cambridge University Press: Cambridge. Cameron, D., White, P., Lampert, R. & Florek, S. (1990). Blowing in the wind: site destruction and site creation at Hawker Lagoon, South Australia. Australian Archaeology 30, 58-69. Camilli, E.L. & Ebert, J.I. (1992). Artifact reuse and recycling in continuous surface distributions and implications for interpreting land use patterns. In J. Rossignol & L.A. Wandsnider (eds) Space, Time and Archaeological Landscapes, pp. 113-136. Plenum: New York. David, B. (1994). The 1984 Chillagoe surveys. Queensland Archaeological Research 10, 37-53. Doleman, W.H. (1992). Geomorphic and behavioral processes affecting site and assemblage formation: the EARM data. In W.H. Doleman, R.C. Chapman, R.L. Stauber & J. Piper (eds) Landscape Archeology in the Southern Tularosa Basin. Vol. 3. Archeological Distributions and Prehistoric Human Ecology. OCA/UNM Report No. 185-324F, pp. 73-105. Office of Contract Archeology, University of New Mexico: Albuquerque. Doleman, W.H. & Stauber, R.L. (1992). Macrotopographic landform analysis as a predictor of archeological remains on the Tularosa Basin floor. In W.H. Doleman, R.C. Chapman, R.L. Stauber & J. Piper (eds) Landscape Archeology in the Southern Tularosa Basin. Vol. 3. Archeological Distributions and Prehistoric Human Ecology. OCA/UNM Report No. 185-324F, pp. 107-151. Office of Contract Archeology, University of New Mexico: Albuquerque.

103

Chapter 3: Spatial Patterning

Fanning, P.C. (1994). Long-term contemporary erosion rates in an arid rangelands environment in western New South Wales, Australia. Journal of Arid Environments 28, 173-187. FANNING, P.C. & HOLDAWAY, S.J. Stone artifact exposure and visibility at open sites in western New South Wales, Australia: a geomorphic framework for survey and analysis. Submitted to Journal of Field Archaeology (received 14th January, 2002). FANNING, P.C. & HOLDAWAY, S.J. (2001a). Stone artifact scatters in western NSW, Australia: geomorphic controls on artifact size and distribution. Geoarchaeology: an International Journal 16(6), 667-686. Gnaden, D. & Holdaway, S. (2000). Understanding observer variation when recording stone artifacts. American Antiquity 65(4), 739-747. Greenwood, S. (1997). Fluvial Processes and Refitting Stone Artefacts at Stud Creek, Sturt National Park. Unpublished B.A. (Hons) thesis, La Trobe University, Melbourne. Hiscock, P. (1985). The need for a taphonomic perspective in stone artefact analysis. Queensland Archaeological Research 2, 82-97. Hiscock, P. (1990). A study in scarlet: taphonomy and inorganic artefacts. In S. Solomon, I. Davidson & D. Watson (eds), Problem Solving in Taphonomy: Archaeological and Palaeontological Studies from Europe, Africa and Oceania, pp. 34-49. Tempus: Archaeology and Material Culture Studies in Anthropology volume 2, Anthropology Museum, the University of Queensland, St Lucia. Hiscock, P. & Hughes, P.J. (1983). One method of recording scatters of stone artefacts during site surveys. Australian Archaeology 17, 87-98. HOLDAWAY, S., WITTER, D., FANNING, P., MUSGRAVE, R., COCHRANE, G., DOLEMAN, T., GREENWOOD, S., PIGDON, D. & REEVES, J. (1998) New approaches to open site spatial archaeology in Sturt National Park, New South Wales, Australia. Archaeology in Oceania 33:1-19. Johnson, D.L. (1990). Biomantle evolution and the redistribution of earth materials and artifacts. Soil Science 149, 84-102. Lancaster, J. (1986). Wind action on stone artifacts: an experiment in site modification. Journal of Field Archaeology 13, 359-363. Linse, A.R. (1993). Geoarchaeological scale and archaeological interpretation: examples from the central Jornada Mogollon. In J.K. Stein & A.R. Linse (eds) Effects of scale on archaeological and geoscientific perspectives, pp. 11-28. Geological Society of America Special Paper 283.

104

Chapter 3: Spatial Patterning

Nash, D.T. & Petraglia, M.D. (1987). Natural Formation Processes and the Archaeological Record. BAR International Series 352. British Archaeological Reports: Oxford. Petraglia, M.D. & Nash, D.T. (1987). The impact of fluvial processes on experimental sites. In D.T. Nash & M.D. Petraglia (eds) Natural Formation Processes and the Archaeological Record, pp. 108-130. BAR International Series 352. British Archaeological Reports: Oxford. Reid, I. & Frostick, L. (1985). Arid zone slopes and their archaeological materials. In A.E. Pitty (ed.) Themes in Geomorphology, pp. 141-157. Croom Helm: Beckenham. Rick, J.W. (1976). Downslope movement and archaeological intrasite spatial analysis. American Antiquity 41,133-144. Robins R.P. 1993. Archaeology and the Currawinya Lakes: Towards a Prehistory of Arid Lands of Southwest Queensland. Unpublished PhD Thesis, Griffith University, Brisbane. Robins, R. (1999). Lessons learnt from a taphonomic study of stone artefact movement in an arid environment. In M.J. Mountain & D. Bowdery (eds) Taphonomy: the Analysis of Processes from Phytoliths to Megafauna. Department of Archaeology and Natural History, Research School of Pacific Studies, Research Papers in Archaeology and Natural History No. 30, pp. 93-106. Australian National University: Canberra. Schick, K.D. (1986). Stone Age Sites in the Making: Experiments in the Formation and Transformation of Archaeological Occurrences. BAR International Series 319. British Archaeological Reports: Oxford. Schick, K.D. (1987). Experimentally-derived criteria for assessing hydrologic disturbance of archaeological sites. In D.T. Nash & M.D. Petraglia (eds) Natural Formation Processes and the Archaeological Record, pp. 86-107. BAR International Series 352. British Archaeological Reports: Oxford. Schiffer, M.B. (1983). Towards identification of site formation processes. American Antiquity 48, 675-706. Schiffer M B (1987). Formation Processes of the Archaeological Record. University of New Mexico Press: Albuquerque. Shackley, M.L. (1978). The behaviour of artifacts as sedimentary particles in a fluviatile environment. Archaeometry 20, 55-61. Seaman, T.J., Doleman, W.H. & Chapman, R.C. (1988) The Border Star 85 Survey: Toward an Archeology of Landscapes. University of New Mexico Office of Contract Archaeology Project No. 185-227. University of New Mexico: Albuquerque.

105

Chapter 3: Spatial Patterning

Stein, J.K. (2001). A review of site formation processes and their relevance to geoarchaeology. In P. Goldberg, V.T. Holliday & C.R. Ferring (eds) Earth Sciences and Archaeology, pp.37-51. Kluwer: New York. Thorley P.B. 1998. Shifting Location, Shifting Scale: a Regional Landscape Approach to the Prehistoric Archaeology of the Palmer River Catchment, Central Australia. Unpublished PhD thesis, Northern Territory University, Darwin. Wandsnider, L.A. (1989). Long-Term Land Use, Formation Processes, and the Structure of the Archaeological Landscape: a Case Study from Southwestern Wyoming. Unpublished PhD thesis, University of New Mexico, Albuquerque. Wandsnider, L.A. & Camilli, E.L. (1992). The character of surface archaeological deposits and its influence on survey accuracy. Journal of Field Archaeology 19, 169-188. Wells, L.E. (2001). A geomorphological approach to reconstructing archaeological settlement patterns based on surficial artefact distribution. In P. Goldberg, V.T. Holliday and C.R. Ferring (eds) Earth Sciences and Archaeology, pp.107-141. Kluwer: New York. Wood, W.R. & Johnson, D.L. (1978). A survey of disturbance processes in archaeological site formation. Advances in Archaeological Method and Theory 1, 315-381.

106

Chapter 3: Spatial Patterning

Stone Artifact Exposure and Visibility at Open Sites in Western New South Wales, Australia: A Geomorphic Framework for Survey and Analysis

Patricia C. Fanning Macquarie University

Sydney, Australia

Simon J. Holdaway University of Auckland Auckland, New Zealand

Submitted to Journal of Field Archaeology 21st December, 2001 (received 14th January 2002).

This article has now been published as:

Fanning, Patricia C. & Holdaway, Simon J. (2002-2004). Artefact visibility at open sites in western New South Wales, Australia. Journal of field archaeology, 29(3/4), 255-271. http://www.jstor.org/stable/3250892 Fanning & Holdaway (submitted) Chapter 3: Spatial Patterning

Abstract

Surface visibility is a significant constraint in archaeological survey, and estimates of surface visibility are a common addition to Cultural Resource Management reports. Despite this, relatively few studies have attempted to identify the factors affecting visibility and quantify their effects. We report the results of such a study based on analysis of surface stone artifacts discarded in prehistory by Aboriginal hunter-gatherers from the Stud Creek area in what is now Sturt National Park, western New South Wales, Australia.

While we are able to demonstrate and quantify relationships between high artifact visibility and erosional surfaces, and low visibility and vegetated or depositional surfaces, our findings also indicate a high degree of local variability. This variability sometimes obscures the predicted relationships. The outcomes of the research lead us to question the way some sampling designs for archaeological survey are constructed.

108 Fanning & Holdaway (submitted) Chapter 3: Spatial Patterning

Introduction

Surface scatters of stone artifacts dominate the archaeological record of the Australian Arid Zone and potentially provide a means of understanding landscape use by Aboriginal people in the past. Survey provides the most common approach to this record (e.g. Thorley 2001, Veth 1993), sometimes based on a siteless or distributional approach (e.g. Robins 1997). Archaeologists seek to draw inferences from the correlation between such variables as site type, artifact assemblage composition or artifact density and environmental factors thought to reflect resource availability in the past. From such data, inferences are drawn concerning the operation of settlement systems in the past (e.g. Ross, Donnelly and Wasson 1992, Smith 1996, Veth 1993). These approaches form the backbone of the Australian Cultural Resource Management (CRM) industry where the goal is significance assessment (e.g. Bowdler 1984, Flood 1984, Hiscock and Mitchell 1993, Pearson and Sullivan 1995), frequently as a response to the threat of destruction of the archaeological record through developments such as mining.

As in other regions of the world, the areas over which the archaeological record is dispersed are frequently too great to permit total coverage survey, therefore a variety of survey designs are employed. Techniques used often involve stratified random designs based on a desire to obtain both estimates of the differential distribution of artifacts across a landscape and the population density of artifacts at any one location (e.g. Thorley 2001).

It is not our intention in this paper to describe or critique sampling methods directly since this is a topic that is well covered in the literature. Instead we focus on the processes responsible for differential artifact visibility, since this is a primary determinant, along with human behavior, of the nature of the archaeological record at the time of the survey. We have previously addressed the effects of post depositional artifact movement on the data collected during artifact survey (Fanning and Holdaway 2001a) and of variability introduced by archaeologists as observers (Gnaden and Holdaway 2000). Here, we consider the degree to which geomorphic processes, the nature of the ground surface, and the presence of vegetation obscure or enhance the ability of archaeologists to observe artifacts.

A number of archaeologists have mentioned surface visibility as a significant influence on the results of archaeological survey (e.g. Hughes and Koettig 1989; McDonald et al. 1994; Smith 1991) but have made these comments in relation to surface vegetation. Grass, leaf

109 Fanning & Holdaway (submitted) Chapter 3: Spatial Patterning litter and trees obscure the surface, making it difficult to see the stone artifacts that form the majority of the archaeological record in Australia. However, lack of vegetation does not automatically ensure that artifacts will be uniformly visible. Surface sediments also obscure artifacts through burial and hence act as an important constraint on estimating artifact density at any particular point in space and ultimately the effectiveness of a particular sampling design. Thus, considerations of visibility need to be extended beyond vegetation to consider the effect of different forms of regolith as well as processes that contribute to the ground surface condition.

In this paper we report on a geomorphological framework for archaeological survey and analysis developed for the Western New South Wales Archaeological Program (WNSWAP) in Sturt National Park near Tibooburra, in arid far northwestern New South Wales (NSW), Australia (Figure 1). The method uses standard techniques of geomorphic landscape mapping to target areas of high artifact exposure. An essential component is an understanding of the geomorphic history of the study area, as well as contemporary landforming processes. Incorporated into our survey design is an assessment of surface condition at a number of different spatial and processual scales discussed below. We present the results of a series of investigations on the effect these processes have on artifact surface visibility.

The Study Area

The rangelands of arid Australia cover 70% of the continent and offer a unique opportunity for archaeologists interested in using surface artifact scatters to study landscape use by Aboriginal people in the past (Holdaway, Fanning and Witter 2000). Since vegetation cover is naturally discontinuous under the prevailing dry climatic conditions, the rangelands are characterized by high levels of artifact visibility. Furthermore, the introduction of European pastoralism in the mid nineteenth century led to accelerated erosion of selected parts of the rangelands, which has enhanced artifact exposure, making the western NSW region an ideal location for development of the framework of archaeological investigation outlined in this paper.

The area chosen for initial investigation was the catchment of Stud Creek, located in Sturt National Park in the far northwestern corner of NSW (Figure 1). It is about 30 km2 in area

110 Fanning & Holdaway (submitted) Chapter 3: Spatial Patterning and underlain by Cretaceous sedimentary rocks dipping gently to the northeast. Silicified units (silcrete) within the sequence form west-facing escarpments and mesas that mark the eastern boundary of the study area. Much of the catchment consists of low-angled slopes mantled with a pavement of silcrete cobbles or 'gibber' (Dury, 1970), while sand patches sometimes abut the base of the west-facing escarpments. The valley floor comprises a relatively thick sequence of alluvial sediments, partially exposed in the walls of the incised channel of Stud Creek. Vegetation cover is sparse, comprising mostly chenopod shrubland, particularly saltbush (e.g. Atriplex vesicaria), with occasional trees such as mulga (Acacia aneura) scattered across the slopes, and gidgee (Acacia cambagei) along the watercourses.

r Cooper Creek iv e R o llo u B

r e

v i

R

o o r a er P iv R n g rli a D

Figure 1: Location of the Stud Creek catchment study area in Sturt National Park, northwestern New South Wales, Australia.

Prior to being incorporated into Sturt National Park in 1972, the area formed part of a pastoral lease for the grazing of sheep and cattle. The introduction of pastoralism in the mid to late nineteenth century led to reduction in vegetation cover and accelerated erosion of topsoil and sediments (Fanning, 1999). Stone artifacts and the remains of heat retainer hearths have been exposed at the surface by this erosion, as the finer topsoil and sediment have been eroded away. The artifacts are mostly located on the valley floor of Stud Creek, adjacent to the contemporary watercourse.

111 Fanning & Holdaway (submitted) Chapter 3: Spatial Patterning

However, while the recent erosional history and the prevailing dry climate means that artifact exposure and visibility is relatively high, it is not complete. Although trees are sparse and mostly restricted to riparian zones, across the rest of the landsurface, low shrubs and grasses grow when and where soil moisture conditions are favorable. As in other areas of Australia, artifacts discarded where the vegetation now grows are obscured. However, contemporary geomorphic processes that have deposited sediments in discrete locations across the landscape, have also acted to cover any artifacts that may be lying on the pre- depositional surface. Animal tracks and various kinds of ant nests are common in the Stud Creek catchment, and indeed across the region. Artifact visibility is therefore also likely to be affected by surface and subsurface faunal activity, through bioturbation and trampling.

Mapping the Geomorphic Landscape to Characterize Differential Artifact Exposure and Visibility

Three aspects of the geomorphic landscape are of particular importance when examining surface scatters of stone artifacts, namely 1 the controls on preservation the archaeological record that are imposed by landscape change over historic, prehistoric and geologic time scales, 2 spatially variable exposure and burial of artefacts due to erosion and/or deposition, and 3 the differential visibility of archaeological materials due to contemporary environmental processes. These landscape attributes have commonly been grouped under the umbrella of natural (as opposed to cultural) formation processes (e.g. Schiffer 1983, Nash and Petraglia 1987), although this division attempts to separate what are in fact the results of two related sets of processes. In the Australian Arid Zone, for instance, the ‘natural’ erosion that has exposed surface archaeological remains has a ‘cultural’ initiation, resulting from over stocking by pastoralists starting in the late 19th century.

Whether thought of as natural or cultural processes, in our study area geomorphic dynamics are the main determinants of artifact exposure, and hence of the proportion of the archaeological record that is available for study at any one time and place. Geomorphic dynamics are reflected in the nature of the landsurface that results from "time-averaging" (after Stern 1994) of the geomorphic environment, i.e. a landsurface is a palimpsest of a set of individual erosional and depositional events operating at a variety of temporal and spatial

112 Fanning & Holdaway (submitted) Chapter 3: Spatial Patterning scales. Averaged over tens to hundreds (and maybe thousands) of years, different parts of the landscape will exhibit a trend to accumulation of sediment (i.e. dominantly depositional), removal of sediment (i.e. dominantly erosional), or no change (i.e. residual). Maximum exposure of the archaeological record is found in those parts of the landscape that are dominantly erosional, while least exposure is found where deposition of sediments is dominant. Therefore, in a distributional approach to the archaeology of surface artifact scatters (Ebert 1992), stratification of the landscape on the basis of dominant geomorphic process allows archaeologists to target surveys where exposure of the archaeological record (though not necessarily its density) is likely to be at a maximum.

The rationale for stratification adopted here is similar to that employed in more recent Australian studies where geomorphological criteria are used (e.g. McDonald et al. 1994, Robins 1997, Thorley 2001). However, our method improves on these by placing weight on the processes that have lead to the preservation, modification and, at times, destruction of the archaeological record as an initial basis for sampling before behavioral inferences are drawn. To some degree it can be contrasted with approaches that rely on ethnographic models as a source of information on which to base survey sampling strategies (e.g. Veth 1993).

Closely related to the identification of depositional, erosional and residual landforms is the question of the identification of the rate, and chronological scale, over which these landforms evolve. While arid regions may present the appearance of an unchanging landscape, the opposite is in fact the case. In our study area, the stratigraphic record dating back to the late Pleistocene is dominated by erosional unconformities (Fanning and Holdaway 2001b). Regional discontinuity in deposition is the norm, leading to a patchwork distribution of landsurfaces differing substantially in age. Rates of deposition or erosion can vary markedly. Some processes take place over tens to hundreds, or even thousands, of years. Others are literally instantaneous. In the Australian summer of 1999/2000, for example, one rain event denuded an area of approximately 7,000 square meters in one of our study areas covered with artifacts and heat retainer hearths. Both artifact and non-artifact clasts were washed into an adjacent gully. This same rain event left other nearby areas relatively unaffected. Such events occurring in the past would effectively reset the archaeological record by wiping out archaeological evidence at a local level.

113 Fanning & Holdaway (submitted) Chapter 3: Spatial Patterning

Understanding the significance of the 'archaeological document', i.e. the proportion of the available archaeological record that is visible to archaeologists at the time of the survey (Wandsnider and Camilli 1992), therefore requires a detailed study of landscape change, like that undertaken at Stud Creek (Fanning and Holdaway 2001b, Holdaway et al. in press). But equally, the archaeological document is a function of contemporary, local scale environmental processes, including erosion and deposition of sediments, bioturbation, and vegetation growth. Current land use practices are also significant, for example whether or not the surface is ploughed, compacted by machinery, or covered by concrete. While also a time average in the sense previously discussed, the contemporary nature of these processes mean that they reflect different temporal and spatial scales to the longer term landscape dynamics discussed above.

Survey Strategy Our survey strategy reflects this temporal and spatial variability in geomorphic dynamics, and characterizes the geomorphic landscape at three distinct spatial scales. At the regional level, the sampling and survey methodology makes use of available Land Systems mapping, a form of regional reconnaissance mapping developed by Christian and Stewart (1953) and applied to extensive tracts of Australia (e.g. Story et al., 1976). Land Systems are defined as “an area or group of areas throughout which there is a recurring pattern of topography, soils and vegetation" (Christian and Stewart, 1953). Land Systems represent a synthesis of the natural environment, and provide, amongst other things, a basis for assessment of land capability and for the study of land use problems (Mabbutt et al. 1973). Land Systems are distinguished on the basis of air photo patterns, and their characteristics are identified by ground observations at sample sites chosen from air photos. Their spatial scale is usually of the order of tens to hundreds of square kilometers.

Since Land Systems are primarily defined by their topographic signature, using remotely sensed data (i.e. airborne and satellite imagery), those geomorphic and other environmental processes that affect artifact exposure and visibility are expected to vary more between Land Systems in a particular region than within any one of them. Therefore Land Systems surveys provide a convenient means of assessing which parts of a region are most likely to exhibit maximum artifact exposure.

At the second or meso-scale level, our survey method focuses on the smaller landform units and landform elements that make up Land Systems, as these reflect the operation of

114 Fanning & Holdaway (submitted) Chapter 3: Spatial Patterning geomorphic processes over the contemporary to historic timescales, which are the most appropriate for surface stone artifact survey. Standard aerial photograph interpretation and field survey methods are used to map landform elements and classify them on the basis of dominant geomorphic environment i.e. residual, transportational/eroding, fully lagged or depositional. In this way, the physical landscape is subdivided on the basis of both landform and dominant processes.

The third or micro-scale level within the survey method concentrates on documenting local variability in landsurface condition that reflects the operation of processes with a short time scale of perhaps hours to days. These landsurface conditions affect the archaeological record at the moment of survey. Such processes include local erosion and deposition of sediments, vegetation growth, and bioturbation.

The method we use to integrate these three survey scales involves the use of a vector GIS (ARCINFO/ARCVIEW) (Holdaway, Fanning and Witter 1997). Each of the three survey scales is mapped as separate polygon coverages in the GIS independent of the distribution of artifacts, which is also mapped as a separate coverage. Data tables hold descriptive information on the nature of each of the polygons. This format permits comparison of the interaction of the three survey scales with the distribution of artifacts. In addition, because of the vector format, data acquisition in the field at all three scales can progress independently.

Implementation at Stud Creek Land Systems maps, at a scale of 1: 250,000, of the whole of the Western Division of NSW, an area of some 32.5 million hectares, are available from the NSW Department of Land and Water Conservation. The Land Systems are grouped on the basis of major landform types, i.e. ranges, tablelands, hillslopes and footslopes; rolling downs and lowlands; alluvial plains, sandplains and dunefields; and playas and basins. The survey area of Stud Creek is contained within the tablelands group, mostly within the Quarry View Land System (Qv), characterized by "stony tablelands with relief to 60 m…escarpments and fringing plains with brown loamy lithosols, moderate mulga, gidgee and belah over bluebush and saltbush" (NSW Soil Conservation Service 1978).

115

Geomorphic Unit Type Geomorphic Unit Name and Relative Area Landform Type Dominant Process(es) Expected Artifact Code Exposure Residual Mesa (RM) Minor Flat-topped, isolated upland Cliff retreat via weathering High remnant capped with silcrete; and block fall bounded by cliffs to 3 m Residual Hill (RH) Minor Rounded low hills below Surface lowering by slow High escarpment mantled with removal of fines via dense stone cover bioturbation and surface wash Lagged Tributary Fan (TFL) Extensive Eroded surfaces of tributary Surface wash of fines by High alluvial fans overland flow; aeolian deflation Footslope (FSL) Moderate Straight low angled slopes Surface wash and aeolian High below escarpment mantled deflation of fines with sparse stone cover Valley Floor (VFL) Extensive Eroded surfaces of main Surface wash and aeolian High valley floor mantled with deflation of fines gravel lag Eroding Valley Margin (EVM) Moderate Severely eroded distal valley Surface wash around domed, High floor margins hardsetting subsoil surface Valley Floor (EVF) Moderate Actively eroding surfaces of Rilling and gullying Moderate main valley floor Tributary Fan Channel (TFC) Minor Channel traversing tributary Concentrated channel flow Dispersed fan Channel Minor Incised distributary channels Concentrated channel flow and None on main valley floor sediment transport; erosion of banks by undercutting and basal sapping Depositional Tributary Fan (TFD) Moderate Depositional surfaces of Deposition of fine sediments Low tributary alluvial fans from surface wash from upslope; aeolian accession Footslope (FSD) Moderate Concave upwards low angled Deposition of fine sediments Low slopes mantled with sediment from surface wash from upslope; aeolian accession Valley Floor (VFD) Moderate Stable surfaces of main Minor aeolian accession of Low valley floor sediments around vegetation Table 1. Geomorphic Units identified and surveyed in the upper Stud Creek catchment.

Fanning & Holdaway (submitted) Chapter 3: Spatial Patterning

Data sets of the upper Stud Creek catchment were constructed largely from intensive field surveys using electronic total stations. The location and extent of artefact sampling surveys are indicated in Figure 2. Several survey base stations were established around the margins of the study area using Australian Map Grid (AMG) co-ordinates with a master station’s coordinates captured using a hand-held GPS. The GPS was left in place for several hours before co-ordinates were recorded, in order to maximize the stability of the signals being received and hence reduce location errors. Once the master base station was established, all surveys were locked in to that co-ordinate system, and digitized aerial photographs, which form the backdrop for the survey data, were geo-referenced using those co-ordinates.

Geomorphic features were surveyed at two different scales (i.e. meso- and micro-) to reflect the different spatial and temporal scales discussed above. At the meso scale, twelve geomorphic units were identified, based on spatial location, topography and dominant process i.e. whether they were residual, actively eroding, fully lagged or depositional (Figure 3 and Table 1). Exposure of artifacts was expected to be greatest on lagged and actively eroding landforms, and the least on those where deposition dominates, such as channel margins on the valley floors (Table 1).

Table 2. Artifact densities on different Geomorphic Units in the Stud 1 artifact survey area.

Geomorphic Unit Area Artifact Number Artifact Density (m2) (n) (n m-2) EVM 2328 6992 3.0 TFL 2595 5751 2.2 FSL 3225 4626 1.4 VFL 1043 1250 1.2 TFD 5321 3389 0.6 FSD 2515 1213 0.5 VFD 7430 330 0.04

Artifacts appeared to be concentrated around clusters of heat-retainer hearths on the valley floors and tributary alluvial fans, and were most clearly exposed on the severely eroded parts of these landforms, for example on the eroding valley margin unit (EVM) and lagged surfaces on the valley floor and tributary alluvial fans (VFL and TFL). Artifacts were less visible on the vegetated slopes and depositional areas of the valley floor. This pattern is confirmed by analysis of artifact density in relation to geomorphic unit from one of the

117 Fanning & Holdaway (submitted) Chapter 3: Spatial Patterning sampling locations, Stud 1 (Figure 2 and Table 2). In this table, the Stud 1 geomorphic units are listed in order of exposure potential from maximum (fully lagged surfaces) to minimum (depositional surfaces). There is a clear correlation with artifact density (Spearman’s rho = 0.89, n = 7, p = 0.007).

Even where erosion is dominant, however, such as on the EVM and VFL units, there are patches of different surface materials which may influence artifact visibility. As discussed above, survey at the level of micromorphology and ground surface condition addresses the problem of differential artifact visibility across the study area at the time of the survey. Features that obscure the ground surface, and hence reduce artifact visibility, include vegetation, plant litter, the presence of topsoil, and sediments of various kinds, especially sand and mud. Other ground surface features both cover artifacts and promote their vertical and lateral dispersal, such as ant nests and animal tracks. On the other hand, ground surface features that promote artifact visibility include bedrock and eroded or lagged surfaces such as subsoil, stone pavements and some kinds of surface crust. Microtopographic features such as rills concentrate surface water flow and hence promote artifact dispersal (Holdaway et al. 1998). Factors that determine the distribution of microtopographic features and ground surface condition within a given area include the operation of contemporary geomorphic processes, local regolith geology, animal behavior, weather and climate, and land use practice. A second (micro-) level of landsurface unit surveying (labeled 'surface condition') was therefore conducted on two portions of the eroding valley margin (EVM) geomorphic unit in the upper Stud Creek catchment. A map of one of these, in the Stud 2 artefact survey area, is shown in Figure 4. As indicated in Table 3, artifact visibility is expected to vary systematically with surface type.

Since artifacts were individually recorded by piece proveniencing over both areas, the relationship between artifact density and ground surface condition could be analyzed. In Table 4, surface condition on the EVM geomorphic unit at Stud 1 and Stud 2 has been listed in order of artifact visibility potential, from highest to lowest. At both Stud 1 and Stud 2, artifact density is positively correlated with surface condition, although this correlation is significant only at Stud 2 (Spearman’s rho at Stud 1 = 0.9, n=4, p = 0.14; Spearman's rho at Stud 2 = 0.9, n = 5, p = 0.04).

118 Fanning & Holdaway (submitted) Chapter 3: Spatial Patterning

Figure 2: Location of artefact survey areas in the upper Stud Creek catchment.

119 Fanning & Holdaway (submitted) Chapter 3: Spatial Patterning

120 Fanning & Holdaway (submitted) Chapter 3: Spatial Patterning

Figure 3: Map of the main valley of the upper Stud Creek catchment, showing the distribution of geomorphic units.

121 Fanning & Holdaway (submitted) Chapter 3: Spatial Patterning

122

Figure 4: Stud 2 artefact survey location, showing distribution of microtopographic features and surface condition types.

Fanning & Holdaway (submitted) Chapter 3: Spatial Patterning

124 Fanning & Holdaway (submitted) Chapter 3: Spatial Patterning

Surface Condition Relative Area Characteristics Expected Artifact Type Visibility Subsoil Moderate Subsoil exposed by High erosion Gravel Moderate Stony (gibber) High pavement Sand/gravel Extensive Mixed sand and Moderate gravel on lagging surface Sand Moderate Medium sediment Low deposition Mud Minor Fine sediment (silt Low and clay) deposition Sand/vegetation Moderate Medium sediment Low deposition around the base of shrubs Vegetation Minor Grasses &/or shrubs Low often growing out of sediment mounds Antnest Minor Circular sediment Low mounds, to 5 cm high and to 50 cm diameter, surrounding nest entrance Rill Minor Linear concentrated Dispersed flow channels across surface

Table 3. Surface condition types and microtopographic features in the Stud 1 and Stud 2 artifact survey areas.

Survey Surface Area Artifact Mean Area Condition Type (m2) Number (n) Density (n m-2) Stud 1 Gravel 366.56 1725 4.71 Sand/gravel 338.80 1775 5.24 Sand 248.29 548 2.21 Mud 19.99 39 1.95 Stud 2 Gravel 16.45 178 10.82 Subsoil 560.96 4111 7.33 Sand/gravel 73.92 441 5.97 Sand 70.50 344 4.88 Vegetation 75.13 26 0.35

Table 4. Artifact densities on different surface condition types in the Stud 1 and Stud 2 artifact survey areas.

125 Fanning & Holdaway (submitted) Chapter 3: Spatial Patterning

Accommodating Differential Artifact Discard Behavior

In the Stud Creek catchment, measured artefact densities confirm the expected relationships between the nature of the land surface, as characterized by landform unit, and the degree of exposure of artifacts lying on that surface (Table 2). Similarly, at the micro-scale level, there is a high level of correlation of surface condition type and the density of artifacts visible on those surfaces (Table 4). However, we would like to be able to use these data to develop a quantitative predictive relationship between artifact density and surface condition, such that if the proportion of surface cover types in a survey area is known then a quantitative estimate of the true artifact density, including those obscured from view, can be made. Unfortunately, this is much more difficult to achieve primarily because the influence of human discard behavior on the distribution of artifacts across the land surface needs to be taken into account.

Artifact discard in the past reflects a variety of behavioral activities; sometimes related to artifact function but including such other activities such as caching, refuse disposal, and abandonment (Wandsnider 1996). The result is a clustered, rather than even or random, dispersal of artifacts across the landscape. Assuming equal visibility, areas with few or no artifacts are interspersed with areas with very high artifact counts as a result of variability in the pattern of artifact discard. Thus, a method is needed for correcting for differential visibility that is sensitive to the underlying pattern of differential artifact distribution.

A further complication arises out of our inability to use excavation as a method of comparison of artifacts visible at the surface at the time of the survey with the available archaeological record (surface and sub-surface artifacts). One of the goals of the WNSWAP research in Sturt National Park was to develop a set of techniques for dealing with surface artifact scatters that did not require excavation. Permission to undertake this work was only given by the Aboriginal traditional owners and the NSW National Parks and Wildlife Service, who manage the national park, on the proviso that no disturbances to the archaeological record, through excavation, take place. Thus, we were unable to compare our estimates of artifact density based on the differential visibility factor with the complete record of surface and subsurface artifacts.

One way to overcome these limitations is to limit the spatial sample over which comparison is made to a small subset of the total area, on the assumption that differential artifact

126 Fanning & Holdaway (submitted) Chapter 3: Spatial Patterning abandonment as a result of human behavior is unlikely to systematically occur over small areas. Small samples can be generated in ARCINFO by using the buffering function, where forming a band a fixed distance either side of a line creates a polygon. In this case, defining a region 2 m to either side of the boundary lines separating adjacent geomorphic unit polygons created a buffer. The area represented by the 2 m buffer in each polygon was then calculated and the INTERSECT function of the ARCINFO software was used to determine how many artifacts rested on each buffer. From these data, relative change in artifact density when moving from one geomorphic unit to the next can be calculated. The results are displayed in Figure 5. There is a significant linear relationship between depositional and lagged surfaces (r2 = 0.54, F= 8.15, p = 0.025) and, as will be shown below, this relationship could be used to predict the influence of differential visibility on changes in artifact density across the upper Stud Creek catchment as a whole.

A similar buffering technique was employed to compare the artifact density of individual surface condition polygons with the density in an area defined by a buffer with a 0.5 m radius from these polygons. The 0.5 m radii reflected the small size of some of the polygons and the desire to limit the area over which comparisons were made so that differential discard behavior could be readily discounted. Where multiple polygons intersected the 0.5 m buffers the CLIP function in ARCINFO was used to remove intersecting areas of the 0.5 m buffer.

At the Stud 2 sampling location the Eroding Valley Margin (EVM) geomorphic unit contains substantial areas of subsoil. On average, the presence of subsoil increases the density of artifacts by 1.73 ± 2.62 artifacts per square meter. As shown in Figure 6, the relationship between the 0.5m buffer artifact density and the subsoil polygon artifact density is linear (r2 = 0.77, F= 76.67 p < 0.001).

For sand, the next most frequent Stud 2 surface condition type, the change is more dramatic with a mean decrease in density of 7.12 ± 4.91 artifacts per square meter when moving from the 0.5 m buffer to the sand polygon. However, the mean value exhibits considerable variability and a plot of the density values for buffer and polygon does not reflect the clear linear relationship seen for subsoil (Figure 7).

Other surface cover types are rare at Stud 2. For gravel, density increases by 8.48 ± 3.08 artifacts per square meter but there are only two areas of gravel at Stud 2 surrounded by

127 Fanning & Holdaway (submitted) Chapter 3: Spatial Patterning eroding valley margin (EVM). For the combination of sand and gravel, there is a decrease of 2.78 ± 4.56 artifacts per square meter (n = 2), while for sand and vegetation, the mean decrease is 2.86 ± 2.25 artifacts per square meter (n = 4).

6 5 4 3 2 1

surface depositional 0 per square meter artifacts 02468 lag surface artifacts per square meter

Figure 5. Relationship between densities of artifacts on lagged surfaces compared with densities of artifacts on depositional surfaces at the Stud 1 artifact survey location. The regression line (r2=0.54, p=0.025) indicates a significant linear increase in density when moving from depositional to lagged surfaces.

25 20 15 10 5 square meter 0 subsoil artifacts per 0 5 10 15 20 25 30 groundsurface artifacts per square meter

Figure 6. Statistically significant relationship between artifact density on subsoil and on adjacent ground surface in the Stud 2 artifact survey location (r2=0.77, p<0.001).

128 Fanning & Holdaway (submitted) Chapter 3: Spatial Patterning

12 10 8 6 meter 4 2

sand artifacts per square sand artifacts 0 5101520 groundsurface artifacts per square meter

Figure 7. Non-significant relationship between artifact density on sand and on adjacent ground surface in the Stud 2 artifact survey location.

18 16 14 12 10 8 meter 6 4 2 0 gravel artifacts per square artifacts gravel 0246810 groundsurface artifacts per square meter

Figure 8. Relationship between artifact density on gravel and on adjacent ground surface at the Stud 1 artifact survey location. Note the two squares with high artifact densities (n = 14 and n = 16 per m2) that strongly influence the regression equation.

129 Fanning & Holdaway (submitted) Chapter 3: Spatial Patterning

At the Stud 1 sampling location subsoil is much less common. There are, however, several gravel polygons that show a mean increase in artifact density of 3.33 ± 4.91 artifacts per square meter. The high standard deviation reflects the presence of two polygons with high densities of artifacts in the 0.5m buffer (Figure 8). If these polygons are removed, the r2 increases from 0.23 to 0.81. Sand/gravel also shows an increase in artifact density but the change (1.60 ± 3.10) is smaller and more variable than for gravel and is not linear.

In summary, it appears that, when differential discard behavior is controlled for, the relationships between surface condition and artifact visibility, as measured by mean density, are not so straightforward. Gravel and subsoil both show a positive linear relationship between the density of artifacts within the surface condition polygon and the area immediately adjacent to it. No simple linear relationship exists for the other surface types present in the Stud 1 and 2 sample areas. Sand and vegetation have a variable, but generally negative, effect on artifact density, although the number of cases is very small. Where sand and gravel or sand and vegetation are found together, moving from the lagged eroding valley margin surface onto the surface condition polygon results in a highly variable change in artifact density.

Towards a Survey Protocol for Accommodating Differential Artifact Visibility

The combination of recording by piece proveniencing and a GIS permitted a detailed assessment of differential visibility for the Stud 1 and Stud 2 sampling locations, but the techniques are not practical for survey across large areas. There is no single, widely accepted approach to assessing site or artifact visibility in the archaeological literature, although it is almost universally recognized as an important determinant of the available archaeological record. Surveyors might include estimates of vegetation cover or density, but often only for descriptive purposes (e.g. DuCros 1990, Seaman et al. 1988, Smith 1991). A more detailed approach to assessing "site" visibility was adopted by Dallas et al. (1995) for an archaeological survey of a proposed power line in far western NSW, Australia, 30 km from our study area, which included cover proportion estimates for various surface types. However there was no attempt to quantitatively relate the cover proportion estimates to the surface artifact visibility estimates. Thorley (1998) attempted a quantitative analysis, but was hampered by the relatively low number of recovered artifacts (420 over a surveyed area of 60,000 square meters). The methods used in Australia for site visibility assessment generally lack the quantitative basis required for realistic estimates of true artefact density.

130 Fanning & Holdaway (submitted) Chapter 3: Spatial Patterning

Therefore, we sought an alternative set of methods that could be applied as part of a sampling strategy employed to study the distribution of artifacts across relatively large areas.

At Stud Creek we adopted a method of quantifying surface condition based on an assessment of cover. In vegetation ecology, crown cover is the most commonly used measure of cover. This is the proportion of a known area within the vertical projection of the periphery of the foliage canopy or crowns. It is visually assessed against standard projections to the nearest 10% (McDonald et al. 1990). For the assessment of surface condition, we have adapted this technique in a similar way to that suggested by McDonald et al. (1990) for determination of the abundance of coarse fragments on the land surface. A visual estimate is made of the proportion of a known area containing or covered by each surface type present in that area, to the nearest 10%. Since cover is two-dimensional, the cover types must not overlap and the total of the percentages of cover types must not exceed 100%. However, this necessitates some special treatment of the data generated, as discussed below. Surface condition assessment is made at the same time as artifact survey.

We laid out a systematic grid of 1m x 1m squares oriented north-south and east-west (Figure 2), and recorded all artifacts in every fifth square. This resulted in 7259 squares being included in the database, a 6% sample of a total area in excess of 120,000 m2. A 6% sampling fraction was selected based on the standard error for the number of tula adzes (Holdaway and Stern, in press) from Stud 1 where all visible artifacts over 20 mm in length were analyzed. The tula was selected for this calculation because it was a relatively rare, yet distinctive, tool type. The 6% sample permitted estimates of the true number of tula at Stud 1 plus or minus 50%. While this error range may seem high, a 6% sample gives estimates of rare types like tula that are of the same order of magnitude as the true number. Estimates of the true number of more common forms are therefore likely to be much more accurate. A systematic, rather than random, sample was applied since the interest was in discovering regions of artifact concentration. Following Wandsnider (1998), the spacing of 1 m squares permits the identification of artifact clusters separated by at least 12 m (i.e. more than 2.5 times the spacing between the squares).

Artifact attributes and surface condition assessment were recorded into palmtop computers in the field using data entry software (McPherron and Holdaway, 1996, 1999) and transferred each night to a relational database and ARCINFO. A unique number assigned to

131 Fanning & Holdaway (submitted) Chapter 3: Spatial Patterning each surveyed square provided a relational link between artifact attributes, surface condition and the count of artifacts in each square. A series of technological and typological variables were described for each artifact, including size measurements (Holdaway, Fanning and Witter 2000). The maximum dimension (a-axis length) of each artifact was measured with calipers to the nearest millimeter. Artifacts shorter than 20 mm were disregarded in order to eliminate the effects of fluvial reworking (Fanning and Holdaway 2001a). Surface condition categories recorded included material types (gravel, sand, mud, topsoil, subsoil, and bedrock), surface crust, vegetation, ant nests, animal tracks and dead wood or fallen timber. Since the method involves subjective assessment of cover percent, maximum consistency was achieved by using just one operator to assess surface condition for the whole of the survey area. Finally, digital images of each square were taken and stored in the database.

Data Analysis: How Mean Artifact Counts Vary With Surface Condition Using SPSS v.10, mean artefact densities were calculated for squares characterized by increasing proportions of each cover type over 50%, and where there was at least one artefact present. For example, the mean artefact densities for all squares exhibiting 50% gravel, 60% gravel, 70% gravel, 80% gravel, 90% gravel and 100% gravel were calculated. The process was repeated for the other cover types. Figures 9 and 10 summarize the results of these analyses. The two figures separate surface types that show markedly different patterns in the density of artifacts as the proportion of each surface cover increases from 50% to 100%. For crust, sand, and vegetation, there is a significant negative linear relationship between the proportion of the square covered and the density of artifacts (Figure 9). For gravel, subsoil and sand/gravel the plots are bi or even tri-modal (Figure 10). Third order polynomials are needed to bring the r2 values above 50%.

132 Fanning & Holdaway (submitted) Chapter 3: Spatial Patterning

10.0 9.0 2 Crust 8.0 Sand 7.0 Veg 6.0 5.0 Crust 4.0 R2 = 0.93 Sand 3.0 Veg 2 2.0 R = 0.84

mean artifact density n/m 2 1.0 R = 0.79 0.0 50 60 70 80 90 100 cover %

Figure 9. Mean artifact densities for squares characterized by increasing proportions of crust, sand and vegetation (veg) where cover is over 50%, and where there is at least one artefact present. Linear least square regression lines have been added to emphasize the relatively uniform decline in mean artefact count as cover proportion increases.

10.0 Gravel 9.0 Subsoil

2 8.0 Sand/gravel

7.0 2 Gravel R = 0.83 6.0 Subsoil

5.0 Sand/gravel

4.0

2 3.0 R = 0.29 R2 = 0.75 mean artifact density n/m density mean artifact 2.0

1.0

0.0 50 60 70 80 90 100 cover %

Figure 10. Mean artifact densities for squares characterized by increasing proportions of gravel, subsoil and sand/gravel where cover is over 50%, and where there is at least one artifact present. Third order polynomial lines have been added to emphasize the variable change in mean artifact count as cover proportion increases.

133 Fanning & Holdaway (submitted) Chapter 3: Spatial Patterning

The non-linear pattern for gravel appears to partially reflect the presence of other low visibility cover types that gravel is most often coupled with, such as vegetation or sand. This was confirmed by examination of various combinations of cover types. For example, in 18 of the 19 squares where gravel cover was 90%, it was found in combination with sand or vegetation, both low visibility surface condition types. Similarly, of the 35 squares where gravel cover was 80%, 28 were in combination with sand, thus reducing the artifact visibility as measured by mean artifact density.

Squares dominated by subsoil are relatively few in number, and in most of them, subsoil is found in combination with sand/gravel, with mean artifact densities varying from 1.43 to 5.09. Sand/gravel is most often found in combination with sand and vegetation, with corresponding low mean artifact densities. High mean densities, around five artifacts per square meter, are found when sand/gravel occurs in combination with gravel, and up to seven in combination with bedrock.

Very limited outcrops of bedrock occurred throughout the study area, and there were very few squares with more than 40% cover of bedrock. Nevertheless, mean artifact densities for squares containing bedrock were generally high, probably reflecting the combination of bedrock with other high visibility cover types, like gravel and subsoil.

The variability, or lack thereof, illustrated in Figures 9 and 10 is also apparent if artifact density and proportion of surface cover are related to the geomorphic unit on which the meter squares rest. All squares were located in the field by a single coordinate in the southwest corner obtained with a total station. These coordinates were converted to a point coverage in the GIS and, using the INTERSECT function of ARCINFO, the geomorphic unit on which the squares fall was determined. Results are presented in Table 5.

Table 5. Mean artifact densities (nm-2) for different surface condition types on different geomorphic units in the upper Stud Creek catchment. Geomorphic unit codes are described in Table 1.

Geomorphic Subsoil Gravel Sand/gravel Sand Vegetation Unit RH None 1.544 ± 1.07 None 1.34 ± 0.60 None FSL 9.57 ± 7.41 4.531 ± 3.81 4.66 ± 4.84 2.58 ± 2.83 1.6 ± 0.69 VFL 2.13 ± 1.13 None 2.30 ± 2.14 1.71 ± 1.24 1.33 ± 0.58 FSD None 2.00 ± 1.58 3.05 ± 3.08 1.82 ± 1.65 1.54 ± 1.0 TFD 3.0 ± 0 3.80 ± 3.11 2.66 ± 2.38 2.15 ± 1.83 1.0 ± 0 VFD None 2.0 ± 0 1.64 ± 0.91 1.57 ± 1.50 1.0 ± 0

134 Fanning & Holdaway (submitted) Chapter 3: Spatial Patterning

All surface types show some variation in artifact density with geomorphic unit, but the variability is highest for subsoil, gravel and sand/gravel. Sand and vegetation show densities that are uniform. Maps of the distribution of the squares help to explain the variability. Squares with sand and vegetation proportions greater than or equal to 50% are distributed in a relatively even manner across the upper Stud Creek catchment. This is not the case for the other surface types. Squares with greater than 50% subsoil are limited to small areas of EVM and VFL. Squares with high proportions of sand/gravel occur differentially on two areas of FSL. Finally, squares with high proportions of gravel are distributed in a complex manner. High proportions of gravel on the residual hill (RH) geomorphic unit show low densities of artifacts, while those on foot slope lag (FSL), particularly the region to the east of the Stud 2 survey location, show high densities of artifacts. Squares located on FSD have intermediate density values. This most likely reflects the fact that stone artifacts were discarded in prehistory in some places in the physical landscape and not in others. The lower footslopes, tributary fans and valley floors were favored locations for the construction of heat retainer hearths, so erosion processes which have exposed subsoil and gravel in these areas are also exposing clusters of discarded artifacts.

In summary, analysis of surface condition on meter squares across a substantial area of the upper Stud Creek catchment indicates variable artifact visibility effects. While there is a consistent decrease in mean artifact density as the proportion of sand and vegetation in the sample squares increases, there is no similar consistent pattern with the other surface condition types encountered in the upper Stud Creek valley. This is because surface cover types like gravel, subsoil and sand/gravel reflect exposure of the original discard pattern via erosion as well as the interaction of surface cover and geomorphic unit. Where samples are relatively small and clustered, as is the case for coarse-grained sediments (gravel, sand/gravel) and subsoil, the result is a complex non-linear pattern. On the other hand, fine- grained sediments (sand, mud) and vegetation obscure the original artifact discard pattern and, if large enough samples over a diverse array of geomorphic units can be obtained, the predicted linear (in these cases inverse) relationship between mean artifact density and visibility (increased cover percent) is apparent.

135 Fanning & Holdaway (submitted) Chapter 3: Spatial Patterning

Discussion

At the beginning of our study we predicted that erosional landforms in the Stud Creek study area (i.e. EVM, FSL, TFL and VFL – Figure 2 and Table 1) would show the highest artifact exposure. The lowest levels of artifact exposure were predicted to occur on depositional landforms (FSD, TFD and VFD – Figure 2 and Table 1). We also expected that gravel and subsoil land surface condition types would promote artifact visibility while sand and vegetation would reduce the number of artifacts observable.

In general these predictions are born out (Tables 2 and 4), but the data are complicated by the differential use of space by people in the past. Despite use of the buffering technique in our analysis to reduce the influence of differential artifact disposal, density changes when moving from lagged to depositional surfaces are variable, although still linear. Greater changes are seen for the smaller surface condition polygons. The general direction of predictions is maintained but on a case-by-case basis there is a great deal of variability, sometimes obscuring the predicted linear relationships.

These results have implications for many of the survey techniques commonly used in Australia and elsewhere. Archaeologists often use narrow transects to sample long linear distances when survey regions are large, or, in the Cultural Resource Management industry, when the development leading to site destruction occurs in a linear form (e.g. pipeline surveys). While our study was not designed to test such survey strategies, our results indicate that differences in artifact visibility caused by relatively localized changes in surface conditions are significant. Even with extensive surveys like ours at Stud Creek, it is difficult to obtain quantitative relationships that would allow the development of a correction factor for differential artifact visibility. This problem is likely to be compounded when samples are taken as long narrow transects. While transects aim to obtain long, linear samples to investigate change in artifact form and density across large regions, the results are dependent on local conditions. As Thorley (1998) remarked on the results of his survey in northern Australia, half the artifacts discovered were found in only 8% of the survey area, a finding largely reflecting differential surface visibility. Despite the desire to sample at a macro scale by using transects, our results suggest that considerable variability will be encountered as a result of changing local conditions. Thus, in some cases, ‘regional level’ differences identified by transect sampling may reflect nothing more than differing local

136 Fanning & Holdaway (submitted) Chapter 3: Spatial Patterning surface conditions between sampling units. Clearly, it is desirable to study the nature of local level variability before assessments of regional significance are made.

Our initial findings based on total artifact provenience (that is, for pieces greater than 20mm in maximum dimension) prompted us to look for a technique that could be applied to the survey of larger areas where piece provenience is impractical. Our solution, the application of surface cover estimates as used in plant ecology surveys, seems feasible in a variety of situations, but our results again emphasize the need for very large samples before interpretations are made. Where sediment and vegetation obscure artifact visibility, we found a linear decrease in artifact density in line with our initial predictions. However, this result may reflect the wide distribution of squares dominated by sand/gravel, sand or vegetation across the upper Stud Creek valley, and their occurrence on a variety of geomorphic unit surfaces. Where the distribution of sampling squares is clustered on separate geomorphic unit surfaces, as is true for gravel and subsoil, the change in density with increasing proportion cover becomes much more complex and is particularly affected by the presence of small areas of depositional materials in sample squares.

This also has implications for survey designs. As noted above, density profiles along transects may be difficult to interpret. The same may be true of density estimates based on sampling units from stratified random designs that seek to correlate artifact disposal patterns with broad environmental zones. While we do not doubt that such correlations exist in many regions, demonstrating the nature of this correlation needs to take into account the difference in artifact density generated by a number of factors operating at different spatial scales. The results from Stud Creek suggest that when large enough samples are taken from one relatively small valley, the predicted relationships hold true. However, smaller samples may be problematic given the level of variability we encountered. We are reluctant to apply our results more generally without additional research. The results of our study suggest that quantifying differential artifact visibility is complex. However, if reproducible, the results from Stud Creek have implications for the Cultural Resource Management industry, where surface assessment of artifact density is a regular practice. We suggest that considerably more effort may be needed to estimate true artifact density where exposure is poor.

137 Fanning & Holdaway (submitted) Chapter 3: Spatial Patterning

Conclusions

Surface visibility has long been recognized as an important factor in assessments of artifact distribution and density. In Australia, where surface artifact scatters dominate the archaeological record, visibility assessment is a regular feature of CRM reports. Increasingly, concerns over visibility as well as other processes that affect the ability to detect a record of past human habitation, have led to the adoption of a geomorphological basis for archaeological survey design. However, despite this approach, there are few instances where artifact visibility is the subject of systematic study.

By considering artifact visibility in the Stud Creek artifact survey area, we are able to demonstrate that erosional surfaces have the highest levels of artifact exposure while artifacts are hardest to find on depositional surfaces. However, our attempts to precisely quantify these relationships reveal that at a local level, there is considerable variation, so much so that at times the predicted relationships between surface condition and visibility are obscured. For the archaeologist interested in studying landscape level relationships in order to reveal the nature of prehistoric settlement patterns, the variability introduced by local differences in visibility needs to be taken into account. Although regional differences in artifact abundance are a common source for behavioral inferences in archaeology, our findings suggest that at times local variation in artifact visibility will obscure these differences or, indeed, create them. Either way, behavioral inferences may be compromised. As Wobst (1983) commented a number of years ago, our ability to understand the past is only as good as our ability to understand the sources of variability that configure the archaeological record.

138 Fanning & Holdaway (submitted) Chapter 3: Spatial Patterning

Acknowledgements

We are grateful to Cecil Ebsworth and family, Wangkumara traditional owners, for permission to undertake the research in their country and for assistance in the field. The project was funded by an Australian Research Council Industry Collaborative Grant to SH while at La Trobe University, and Macquarie University Research Grants to PF. We thank Damian Gore, Judith Littleton, John Pickard, Justin Shiner and LuAnn Wandsnider for helpful comments on early drafts.

Authors’ Biography

Patricia Fanning is a Senior Lecturer in environmental science and management in the Graduate School of the Environment at Macquarie University, where she has been teaching since 1990. Initially trained as a geomorphologist specializing in erosion processes and landscape change in arid Australia, her interests have extended to include geoarchaeology and cultural heritage management. Mailing Address: Graduate School of the Environment, Macquarie University, NSW 2109, Australia.

Simon Holdaway is a Senior Lecturer in the Department of Anthropology at the University of Auckland. He received his PhD from the University of Pennsylvania (1991) dealing with the early Upper Paleolithic of France. Since that time he has worked in Australia on the Holocene prehistory of the arid zone and the Pleistocene prehistory of Tasmania. He has coauthored, with Nicola Stern, a book on Australian prehistoric stone artifacts (Holdaway and Stern in press). Mailing Address: Department of Anthropology, The University of Auckland, Private Bag 92019, Auckland 1, New Zealand.

139 Fanning & Holdaway (submitted) Chapter 3: Spatial Patterning

Bibliography

Bowdler, Sandra 1984 “Archaeological significance as a mutable quality”, in Sharon Sullivan and Sandra Bowdler, eds, Site Surveys and Significnace Assessment in Australian Archaeology. Canberra: Department of Prehistory, Research School of Pacific Studies, The Australian National University, No. 13: 1-9. Christian, C.S. and Stewart, G.A. 1953 General Report on Survey of Katherine-Darwin Region, 1946. Commonwealth Scientific and Industrial Research Organisation, Australian Land Research Service 1. Canberra: C.S.I.R.O. Dallas, Mary, Peter Mitchell, Huw Barton and Jeremy Beckett 1995 “Archaeological and Anthropological Study of the Proposed Connection of Grid Power to Sturt and Mootwingee National Parks Environmental Impact Statement”, unpublished report to Manidis Roberts Consultants on behalf of the NSW National Parks and Wildlife Service. Sydney: Mary Dallas Consulting Archaeologists. Du Cros, Hilary 1990 The Sydenham Corridor: A Cultural Heritage Study. Victoria Archaeological Survey Occasional Report 26. Melbourne: V.A.S. Dury, George H. 1970 “Morphometry of gibber gravel at Mt Sturt, New South Wales”, Journal of the Geological Society of Australia 16: 655-666. Ebert, James I. 1992 Distributional Archaeology. Albuquerque: University of New Mexico Press. Fanning, Patricia C. 1999 “Recent Landscape History in Arid Western New South Wales, Australia: A Model for Regional Change”. Geomorphology 29: 191-209. Fanning, Patricia C. and Simon J. Holdaway 2001a “Stone Artifact Scatters in Western NSW, Australia: Geomorphic Controls on Artifact Size and Distribution”. Geoarchaeology: an International Journal 16(6): 667-686. Fanning, Patricia C. and Simon J. Holdaway 2001b “Temporal Limits to the Archaeological Record in Arid Western NSW, Australia: Lessons from OSL and Radiocarbon Dating of Hearths and Sediments”, in Martin Jones and Peter Sheppard, eds, Australasian Connections and New Directions, Proceedings of the 7th Australasian Archaeometry Conference. Research in Anthropology and Linguistics 5, 85- 104. Department of Anthropology, University of Auckland: Auckland.

140 Fanning & Holdaway (submitted) Chapter 3: Spatial Patterning

Flood, Josephine 1984 “More or less significant: a national perspective on assessing the significance of archaeological sites”, in Sharon Sullivan and Sandra Bowdler, eds, Site Surveys and Significance Assessment in Australian Archaeology. Department of Prehistory, Research School of Pacific Studies Publication 13: 55-60. Canberra: A.N.U. Gnaden, Denis and Simon J. Holdaway 2000 “Understanding Observer Variation When Recording Stone Artifacts”, American Antiquity 65(4): 739-748. Hiscock, Peter and Scott Mitchell 1993 Stone Artefact Quarries and Reduction Sites in Australia. Australian Heritage Commission Technical Publication Series 4. Canberra: A.H.C. Holdaway Simon J., Patricia C. Fanning and Dan C. Witter 1997 “GIS Analysis of Artefact Distributions on an Eroding Landscape: The Western New South Wales Archaeological Project”, in Ian Johnson and MacLaren North, eds, Archaeological Applications of GIS, Proceedings of Colloquium II, UISPP XIIIth Congress, Forli, Italy, September 1996, Sydney University Archaeological Methods Series 5. Sydney: Archaeology,University of Sydney. Holdaway, Simon J., Patricia C. Fanning and Dan C. Witter 2000 “Prehistoric Aboriginal Occupation of the Rangelands: Interpreting the Surface Archaeological Record of Far Western New South Wales, Australia”, The Rangelands Journal 22: 58-71. Holdaway, Simon J., Patricia C. Fanning, Dan C. Witter, Martin Jones, Geoff Nicholls and Justin Shiner in press “Variability in the Chronology of Late Holocene Occupation on the Arid Margin of Southeastern Australia”, Journal of Archaeological Science. Holdaway, Simon J. and Nicola Stern in press Written in Stone. Melbourne: Museum Victoria Press. Holdaway, Simon J., Dan C. Witter, Patricia C. Fanning, Robert Musgrave, Grant Cochrane, Trudy Doelman, Simon Greenwood, Daniel Pigdon and Jamie Reeves 1998 “New Approaches to Open Site Spatial Archaeology in Sturt National Park, New South Wales, Australia”, Archaeology in Oceania 33: 1-19. Hughes, Phillip J. and Margaret Koettig 1989 “Recording methods: a consultant's view”, in Josephine Flood, Ian Johnson and Su Sullivan, eds, Sites and Bytes: Recording Aboriginal Placies in Australia. Australian Heritage Commission Special Australian Heritage Publication Series 8. Canberra: A.H.C., 3-19. McDonald, Josephine, Elizabeth Rich and Huw Barton 1994 “The Rouse Hill Infrastructure Project (Stage 1) on the Cumberland Plain, Western Sydney: Recent Research and Issues”, in Marjorie Sullivan, S. Brockwell and A. Webb, eds,

141 Fanning & Holdaway (submitted) Chapter 3: Spatial Patterning

Archaeology in the North. Darwin: North Australia Research Unit, Australian National University, 259-293. McDonald, R.C., R.F. Isbell, J.G. Speight, J. Walker and M.S. Hopkins 1990 Australian Soil and Land Survey Field Handbook, 2nd Edition. Melbourne: Inkata. McPherron, Shannon and Simon J. Holdaway 1996 “Entrer Trois”, in Harold Dibble and Shannon McPherron, eds, A Multimedia Companion to The Middle Paleolithic Site of Combe-Capelle Bas (France). Philadelphia: The University Museum, CD-ROM. McPherron, Shannon P. and Simon J. Holdaway 1999 “Entrer Trois”, in Juan A. Barceló, Ivan Briz, and Assumpció Vila, eds, New Techniques for Old Times: Computer Applications and Quantitative Methods in Archaeology. Bar International Series 757. Oxford: B.A.R. Mabbutt, Jack A., Juliet P. Burrell, Janice R. Corbett and Marjorie E. Sullivan 1973 “Land Systems of Fowlers Gap Station”, in Jack A. Mabbutt, ed., Lands of Fowlers Gap Station. Fowlers Gap Arid Zone Research Station Research Series 3. Sydney: U.N.S.W., 25- 45. New South Wales Soil Conservation Service 1978 “Milparinka”, Land Systems Series Sheet SH 54-9. Sydney: NSW Government Printer. Nash, D.T. and M.D. Petraglia 1987 Natural Formation Processes and the Archaeological Record. BAR International Series 352. Oxford: B.A.R. Pearson, Michael and Sharon Sullivan 1995 Looking After Heritage Places. Carlton: Melbourne University Press. Robins, Richard P. 1993 “Archaeology and the Currawinya Lakes: Towards a Prehistory of Arid Lands of Southwest Queensland”, unpublished Ph.D. thesis, Griffith University, Brisbane. Robins, Richard P. 1997 “Patterns in the Landscape: a Case Study in Nonsite Archaeology from Southwest Queensland”, Memoirs of the Queensland Museum, Cultural Heritage Series 1(1): 23-56. Ross, Anne, Terrence Donnelly and Robert Wasson 1992 “The Peopling of the Arid Zone: Human-Environment Interactions”, in John Dodson, ed., The Naive Lands: Prehistory and Environmental Change in Australia and the Southwest Pacific. Melbourne: Longman, 76-114. Schiffer, Michael B. 1983 “Toward the Identification of Formation Processes”, American Antiquity 48(4): 675-706. Seaman, Timothy J., William H. Doleman and Richard C. Chapman 1988 “The Border Star 85 Survey: Toward an Archeology of Landscapes”. Office of Contract Archeology, University of New Mexico, Project 185-227. Albequerque: UNM.

142 Fanning & Holdaway (submitted) Chapter 3: Spatial Patterning

Smith, Laurajane 1991 “Berwick-Pakenham Corridor Aboriginal Archaeology”. Victoria Archaeological Survey Occasional Report 45. Melbourne: V.A.C. Smith, Michael A. 1996 “Prehistory and Human Ecology in Central Australia: An Archaeological Perspective”, in Steve R. Morton and D. John Mulvaney, eds, Exploring central Australia: Society, the Environment and the 1894 Horn Expedition. Sydney: Surrey Beatty and Sons, 61-73. Stern, Nicola 1994 “The Implications of Time-Averaging for Reconstructing the Land-Use Patterns of Early Tool Using Hominids”, Journal of Human Evolution 27: 89-105. Story, R., R.W. Galloway, J.R. McAlpine, J.M. Aldrick and M.A.J. Williams 1976 “Lands of the Alligator River Area, Northern Territory”. C.S.I.R.O. Land Research Series 38. Thorley, Peter B. 1998 “Shifting location, shifting scale: a regional landscape approach to the prehistoric archaeology of the Palmer River catchment, central Australia”, unpublished Ph.D. thesis, Northern Territory University, Darwin. Thorley, Peter 2001 “Uncertain Supplies: Water Availability and Regional Archaeological Structure in the Palmer River Catchment, Central Australia”, Archaeology in Oceania 36: 1-14. Veth, Peter Marius 1993 Islands in the Interior: The Dynamics of Prehistoric Adaptations within the Arid Zone of Australia. International Monographs in Prehistory, Archaeological Series 3. Ann Arbor: I.M.P. Wandsnider, LuAnn 1996 “Describing and comparing archaeological spatial structures”, Journal of Archaeological Method and Theory 3: 319-384. Wandsnider, LuAnn 1998 “Landscape element configuration, lifespace, and occupation history: ethnoarchaeological observations and archaeological implications” in A. P. Sullivan III, ed., Surface Archaeology. Albuquerque: University of New Mexico Press. 21-39. Wandsnider, LuAnn and Eileen L. Camilli 1992 “The Character of Surface Archaeological Deposits and its Influence on Survey Accuracy”, Journal of Field Archaeology 19: 169-188. Wobst, Martin H. 1983 “We Can’t See the Forest for the Trees: Sampling and the Shapes of Archaeological Distributions”, in James A. Moore and Arthur S. Keene, eds, Archaeological Hammers and Theories. New York: Academic Press, 37-8

143 144

Publication

Due to copyright laws, the following article has been omitted from this thesis. Please refer to the following link for the abstract details.

Fanning, Patricia & Holdaway, Simon (2001). Stone artifact scatters in western NSW, Australia: geomorphic controls on artifact size and distribution. Geoarchaeology, 16(6), 667-686.

http://dx.doi.org/10.1002/gea.1015

145

146

147

148

149

150

151

152

153

154

155

156

157

158

159

160

161

162

163

164